Document KRg6eDYdQ57raaXn5aEw9YRb0

Recorded telephone conversation between Dr. J. K. Wagoner & Ms. Selevan of NIOSH and Dr. J. Rafter of & Research - Jan. 4 Sherry 1977 J.R. SS J. $ J.R. JKW J.R. JKW J.R. JKW Let me tell you what I have then we'll kind of be on the same wave length maybe I have transcript that Ms. Selevan gave at the American Public Health Association on October 20th I have a transcript of that talk I also have a listing of th 94 deceased individuals and a more detailed listing of 17 who died from respiratory causes and finally I have some tables which I understand were given to the three companies involved in your probably equivalent to the report slides and I think they're that were shown on the they are equivalent Pardon 7 Fra who ? So that's where I stand and now I need generally some information so I can understand how they were analyzed and maybe go over it myself so I can report to people that ask me whether or not I think it's a reasonable analysis and so forth Well basically the underlying assumption of the whole analyses are out with ten ten persons that one person person years at observed for one observed risk which is to equates out equates say that to ten person years at risk However there is one additional parameter that added into it because of the fact that disease causation if not an instantaneous event as manifest by death draw out and that is we analyzed our on a piece of paper a matrix basic which data if you has on one could scale axis duration since onset of which for easier terminology we classify as latency on the you have duration of employment Within out seven different categories in each one other axis if you draw of those axis total actually which is eight being the marginal cell'1 x 1 you with is the first category less than five years of duration of employment and less than five years since onset of employment Going over to your right the second cell is 5 through 19..where have you got duration of employment on top on top Within latency less than five 5 to 9 years it's a nonsense You understand that years duration of employment cell it's an impossibility o.k. Yes Basically let me go back one further you the five variables six variables step I'll give that we consider We consider age whicihs an always increasing function Protected Document to Protective order 0148405 JNJ 000042669 JKW JR JKW J.R. JKW -2- we consider calendar time which is an always function we consider interval since onset of which is an always increasing function _. increasing employment that's latency is that correct latency and we consider duration employment which is a static or increasing function depending on whether a person terminates employment or not Those are your basic four variables that are considered other than W in variables such as race and sex Now within each one of these cells in this big matrix that I had given we have a life table in terms of calendar time and you in five year intervals calendar time in age age five year intervals The reason that we analyze in terms of these two variables is that disease is obviously associated with is also a secular phenomenon terms of aging disease and things like this that diagnostic differences lung cancer is an increasing function over time and what we hope to do when we analyze is to control for these nonsense variables so that later in our final analyses we can after analyzing for these variables we can collapse the table and just consider dosage and intervals since onset holding each of these other variables constant as we controlled for them in the analyses O.K Yes On your axis you have five year intervals less than five five to nine ten through fourteen fifteen through nineteen twenty through twenty twenty through twenty greater equal to thirty and a total which is the total summation of all the events in that column r laorwgestIn going down you have the same numeric breakout of less than five five through nine O.K. What we do is we take each individual as he comes in we ask the basic did he get employment into the life table in he began employment 1940 the question in 1940 entered age he had achieved in 1940 latency 0 because he had just began employment and the duration of onset of employment of O. that he has continuous Assuming he the employment passes down through diagonal of the table both in terms of calendar time in terms of aging in terms of in terms of interval since onsetinocfreaesmiplnogymeemnptlouynmteinlt heand stops employment in which case he no longer increases in terms of duration of employment but just increases in terms of onset interval since onset of O.K. As he ages of course he employment He can get out of the life tableproingrets ersmess oifn c tw aolenddiafrferteinmte events Either he expired in which case he is from risk of dying at the date that he removed expired and is removed totally from further accumulation of years at risk of dying at that point in time ro irskhe pe ca rn son come to the end of the study period when out of the everybody drops study who does not expire O.K. Now if a continues employment from the date he began employmentperisnon 1940 and the study period ceases in 1975 he will contribute he five years of risk in cell will contribute five of risk in cell II O.K because he has spent five yea yr es ars Protective Protected Document to Protective order 0148406 JNJ 000042670 JR JKW JR JKW at risk of dying as he was accumulating his fifth through his ninth year of employment and he was in his fifth through ninth year since onset of employment at that time And like- wise five in the third cell 33 five in 55 five in cell 66 and six in cell 77 '73 '74 and '75 O.K. Now let's assume cell that that 44 five in cell is 1970 '72 we have another individual that comes through but he dies on the 10th anniversary of his initial employment He will contribute five years at risk in cell II he will contribute five years at risk in cell 22 and will be removed from the life table at that point in time as deceased due to the underlying cause as certified on the death certificate O.K. I will take a third individual who just dies on the fifteen anniversary of his employment initial employment but he only works for ten consecutive | years He will in cell 22 and of employment contribute five years five years in cell II five years in cell 23 two being duration Its 32 across the way the top that I have it would be it I 32 in guess since I terms of o.k. have duration Essentially he goes on the diagonal and then drops one cell and then drops out due to the cause of death that he died from O.KJ You with me Yes The only other case that really theoretically can happen you can work out all sorts of combinations on those would be a person who was initially employed worked for five years left the industry for five years came back and worked for another five years Alright That person contributes five years at risk in cell in cell 12 one being at the fifth to ninth HI he contributes five years at risk less than five years of duration but year since onset o.k. and then he contributes five years at risk in cell 23 during the period that he accumulated his fifth through ninth year of employment but he was in his tenth through his fourteenth years since onset of employment That will be third type of individual fourth type of individual that can fit into the life table at person who begins risk The employment other condition obviously a in 1920 of our study period record begins in 1940 because that when the when our follow up started because we system or thats social security as method of ascertaining whether a person is deceased or alive and that system really came into effect in 1935 oractually '37 or '38 If that person began in 1920 and was still on the payroll in 1940 then this is the first date of our study period He will be in cell 5-5 when he enters our period at risk He has accumulated 20 through 24 years of 20th year of the onset employment he's actually on the of his employment He contributes no risk during the interval prior to that because he has no probability of dying otherwise he would not be on the payroll of 1940 when we had the first chance of picking Protected Document to Protective order 0148407 JNJ 000042671 R. JKW JKW J.R. him up O.K Yes Basically from our you have the in this case matrix as we send people through the Vermont Talc workers themselves Each one of these of these obviously as they go through each one cells is put into that additional matrix within the eight periods through '50 1940 cell of a series of sixteen age groups and approximately calendar 1945 through 1949 through 54. O.K. These individual numbers that we are distributing are individual person years that each individual contributes that risk of dying during the period of the study O.K Now to get what one would expect to have occurred in this population if they had experienced the same death rate of the control group this case the U.S. calendar male period population we generate out for the same that we used in our life table distribution and the same age groups that we use in our life table distribution death per 100,000 population due to specific causes of death 001 through 0119 ICD code tuberculosis malignant - 140 through 199 or 205 comparbilty In order to provide comparability comparability because the coding of death study has changed over deaths in terms of period of time we interpret our the revision that was in effect at the time of death and then we convert that code to a Sc o o wemdp o haavreability seventh revision code number for ease of handling O.K. and death statistics as Statistics Division and We code out the U.S. population published by the National Vital put them together using that same comparability arrangement We take these death rates for 100,000 population specific for five years age groups and five year calendar time periods and bring them back in and multiply them by the person years at risk in the life table that we just distributed which are now specific | for calendar time and age if you -and for ease group within dosage and of statistical notation latency we will LYZ abcd say that lets make it where abcd a is dose 1 b is latency ^ is age group 1 and d is calendar time 1 we take that person year at risk times the death rate of the general population specific for age group c and calendar time group b O.K. That generates out the expected numbers of deaths that would occur in this population within age group c calendar time period time d experienced group 1 latency group 1 if they had experienced the death rate due to that particular cause of the general population K. Let me ask you a question references for the United what was it sex at this point You use these States and broke them down by ) sex race cause age calendar time Oh so there is age in there Protected Document to Protective order 0148408 JNJ 000042672 JKW JR JKW SS JR SS J.R. JKW JR JKW JR JKW JR JKW Oh yes Age is in there sex is in there race is in there These are race sex specific race How about job classifications in these Is there anything like that Well not at this time here O.K. clatification at a later time We'll get into that sub You mean for the U.S. rate Yeh I'm just trying to get an idea of what these references give The rates for the U.S. don't have any I see recupation's information These are just U.S. general white Yet anything the use of the U.S. male population rates white male population general population as a reference population tends to underestimate what one would expect in all things other than age and sex rate a populatioinf and calendar time were the same The rate in being that a working population consists of people who are physically able enough to seek employment and are medically sound enough to pass routine company medical examinations in order accepted for employ- ment They are also well enough to withstand the normal rigors of a physical type of employment whereas the general population not only includes many of the same people from that industry but all other industries but they also contain the sickly the people who are in hospitals with diagnosed cancer already who have yet to expire and things like that In addition during the war years you tend to find that the general medical population rejects and in many cases contains many of things like this which are left the in the U.S. population and during those years the population at risk really was continental U.S. not population overseas etc. The use of the general population has been referenced in many articles as producing what we call the healthy worker effect and that is in underestimate of the risk that than one would experience if these nonsense variables in fact everything were equated other Is this information reference available to me in the library someplace What the healthy worker effect No the references that you used for the U.S. Yes they're published vital statistics That's what I would ask for Each year they put out two big volumes on it Protected Document to Protective order 0148409 JNJ 000042673 JR JKW JR JKW JR JKW JR JKW JR JKW JR JKW JR Who puts it out Public Health The U.S. Vital Statistics Division What about the references available someplace for Vermont Are they also There are two different types of references Either you can get population and death from the same volume that we were talking about causes of death as for you many but can from not the as large an array of U.S. The U.S. breaks out about 251 specific causes of death I think each state breaks 68 or 56. Its a less detailed breakout in terms of specific types of causes of death or the state of Vermont themselves have their own data available I would have to contact an agency in Vermont I would imagine the State Health Department State Health Department and this this state Health Department also has job classifications No no . job classifications The reason being that one assumes that the general population as I said before if anything underestimaant d etoso that it contains because of the fact that its a large population relative to a very small restrictive geographic region tends to contain all segments of the population and is really in sense smoothed out in terms of risk factors O.K. Yes I understand When one gets through with this whole array that we are talking about where one increments the calendar time and increments the age and then goes in and calculates another expected death and goes all the way through for each one of the cells then we update the cause of death and go into the second cause of death and do all the same sequentially through the ek. When all of this is done we sum over all calendar periods and all age groups within specific latency and dosage cell O.K controlled for age the effects The reason of age we being we have now have standardized for the effects of age and calendar time and we are really causation not interested in terms of these as disease standardized for and held at constant We summarize over those and then within latency You with me we sum within dosage over latency and we sum over dosage to get our marginal total O.K. a Yes Then come you out can the sum your marginal total and they same O.K which will give you a had better grand total Right of the number of deaths Protected Document to Protective order 0148410 JNJ 000042674 JKW JR JKW SS JKW JR JKW Number of deaths both observed number of deaths expected and you can sum your person years because they're additive also Only for descriptive purposes the actual years of the marginal total person years or the grand total person years in any calculation is statiscally in balance because you in a sense have sacrificed your age calendar time distribution that consideration O.K. You with me the general philosphy behind the Well basically that is people were presented by Ms. Selevan on occasion presented the grand total for various causes of death on other occasion presented the sub marginal totals according to intervals since onset of employment as we wish to look at latency O.K. The reason we wish to look at latency is if in fact one usually sees or one sees that one has an excessive risk as measured by the ratio of the observed over the expected times 100 which is the SMR early in the game with very short latency one can become very suspicious of an occupational ideology toward that excessive risk The reason being that its impossible or very unlikely that one has instantaneous death due to an exposure for a chronic disease Of course there is a problem there and that is in some instances people were exposed a very short time a long time ago and it appears that they have a long latency where you know they have a long time since exposure but they had a very small exposure tends to be misleading in some cases I would think Well you've got your option in looking at both spectrums dosage and latency O.K. The original tables that you've got they just looked at latency because the effect was much more pronounced there It was a ten minute presentatioInn the final draft we also looked at exposure but that opportunity we just didn't have enough time to give all the data and the pro- nounced effect was with latency a much lesser effect as there Well was one with exposure ee eee ? Ae tk can normally say is we that we pagon a hat hat wala wala din din really don't know summ how much duration of employment it takes to really induce a disease either it manifests by its diagnosis and death but if its a chronic disease after a person was exposed in terms of diagnosis and death is very minimal Certainly all of our population studies to date have de- monstrated that I'm sorry that I didn't quite understand that would you repeat What I'm saying is that whereas your dose group 1 in your latency group would be not expected to have an SMR above 100 o.k. the ratio of observed over expected should not even exceed 1 scale latency the further you get down removed on the with the possibility of certain exposure Protected Document to Protective order 0148411 JNJ 000042675 -8- oo JR JKW JR JKW JR JKW JR JKW JR JKW if there is an affect you would expect to see an increasing relative risk Ratio of observed to expected If in fact there is an occupational would increase ideology if in fact short term exposure is really associated with this phenomena O.K. There is another question that occurs to me and that is since there you have very few people in your sample that you are working with does this change what you've just said for example I can understand that if things would even out kind in few people can you a lot ..can of people but when mistakes be made by you have very making that assumption because you just have so few people Well I think one could hood of overestimating basis that the test of distribution take into numbers tend to underestimate but the likeli- the risk would be unlikely on the significance uses a Foisson ? consideration the magnitude of the Does this test of significance something that you do or is it done by your computer of Really we use table Biometrica Volume I Biometrica Table for Statisticians Persall & Hartley Its a standard process that is used internationally in the field So given an SMR then you would look in this table Actually the table gives you the bounds on the observed number of cases I see I noticed in one place in these tables you made comparisons between the expected number for the Vermont miners and Vermont population excuse me the observed for the miners and the expected for the Vermont as well as the observed for the miners and the expected for the U. S. population We use two different standard populations to generate our expected one the U.S. white male population and two the Vermont white male population Did you use the same table to do the analysis and does the table take into account that you are doing multiple comparisons using the same expected number of people the same observed number of people No no Protected Document to Protective order 0148412 JNJ 000042676 JR So its like you had several tests to do and you just did pair comparisons with each of the possible JKW Right right JR Have you adjusted your P values for that at all JKW nept No no In this case I would doubt very much if it would make any difference in light of the extremely large relative risk And this is going to be heresy that I'm going to state being a graduate of Harvard University School of Public Health both in biostatistics and epid^'- miology but I often view tests of significance to be a crutch that have been imposed upon the statisticians by the incompetent medical profession who are too damm inadept to interpret their own data I would prefer to biologically interpret the significance of my data as well as the statistical significance of my data JR I should hope anyone would do that JKW Too often this is not the case JR JKW JR JKW JR JKW JR The unfortunate thing is people will P values and take them as gospel I with you tend to think I look at am agreeing This basically is why we tend to restrict ourselves to just the .05 and anytime you start getting down to the exact and multiple comparisons its ludicrous Of course times does the sample the fact that you reuse the same sample several tend to make the conclusions more dependent upon and less relevant to the total population It depends on which standard population you are referring to Well I assume that the underlying population is something along the lines of people who are or will be employed in talc mines that's the inference I guess that's being No. I mean in which general population If one does multiple comparisons and looks at the Vermont population and says let's look at the U.S. population certainly the first is an in- dependent step Yes that's true expect you're using the same observed in each case that was my point Could I ask you some specific questions about the people that had died of respiratory causes For example on this listing there are about ten of them where its listed as smoking unknown has this been updated at all is there new information about that Protected Document to Protective order 0148413 JNJ 000042677 -10- JKW JR JKW SS JKW JR JKW SS JR I think there is update going on on it yes Obviously we are attempting to get the complete smoking history as very possible However I think one has to I would hope that we are not going to be using categorically smoking as red herring and let me explain whatI mean on that Single population of approximately 60 x or current smokers that We are not using U.S. white male population rates specifically for smokers when one tends to say well you haven't considered cigarette smoking and obviously it must be due to cigarette smoking it is often stated that we would without consideration being given to the type that a sizable portion of the general white male population are also current and x cigarette smokers In spite of that fact that we do have smokers vs. smokers and they may cigarette smoking but there is no that at present differ in quantity of data which would support The only criticism . I would have of that is that its generally true if you have a large sample again the fact that you have a rather small sample cigarette smoking could be more of an influence than if you had a larger sample even though as you mentioned your Pason tables take into account small numbers Tel Sy There are a few other things that I think we ought to consider Since we break the population down into miners and into millers two independent categories and miscellaneous Miscellaneous are mixed categories both places people who worked in O.K. Yes I know about that The likelihood of an independent effect of cigarette smoking in the ideology this diseas that we are observing seems very small unless one further evokes the hypothesis that there is job specific smoking patterns within specific corporations in the same geographic area That is why do miners smoke less than millers and if they don't why do we see an excessive risk of malignant respiratory disease in the millers and not the miners Why also cancer in conversely the miners do we see an excess of but not the millers respiratory Well the point is I agree that you can make you can ask these questions except when you've got only like for example people who died of cancer and if two or three of them died as a .. which I don't know but if two or three died as a result of cigarette smoking you know that really affects the sample If you had sixty people and two or three had died as a result that doesn't effect the sample Protected Document to Protective order 0148414 JNJ 000042678 C ; -11- JKW SS JR SS JR JKW SS JR JKW JR JKW SS JR SS JKW SS JKW SS | We're not pushing on the lung cancer at the present time Mie We are not making any conclusions its just that you work would expect consistent patterns if smoking is truly the cause of the excessives we're observing you would expect the pattern not to be opposite but to be consistent and they are That's the only point I'm making I'm not drawing any conclusions as to the lung cancer at this point /] ff / My point is that its hard to see trends when you have few people Well the only trend that we're really is that and respiratory diseasIet is specifically striking identifying Let me ask you do you have information with regard to length of exposure and period of employment of the people other than the ninety four who died I understand there was a sample of 405 taken Do you have information on these people who are still alive with regard to things like exposure to tale and working history oh yes They are distributed in the matrix This includes everyone living and dead in that 405 people . And they are included in the calculation of expected 7 on yes In order to generate death you have to have people in years and who observe and expected are generatinogf course. , All of these that adds up people add up to to more than 94. only have 94 there's none of them I took it to mean that you Oh no there are 94 deaths Actually there's a correction there are people without death certificates and we two of those 94 to be alive But its 92 corrected tables 92 we had some had since found now we will have printints of Is it possible for me to get notes the living people with regard to length of exposure That information is confidential It will be made available What the living people No No No. He asked for living people of person years You want distribution ; Protected Document to Protective order 0148415 JNJ 000042679 -12- JR SS JR SS JR JKW JR SS JR SS JR JKW Well what I would like is that which gives me a distribution as exposure would be helpful something to the regard to the You don't workers want actual individual information about individual No I don't need regard to length of employment of history of those that I'm interested in distributions with of exposure of live live people and if live people people and also period possible the smoking 1 ( We don't have smoking histories of live people don't people Is it possible for me to get that or can I come and pick it up or can you send it to me Can I ask another question while you are deciding First of all we do not have cigarette smoking habits patterns among the surviving O.K. Many of the surviving people no longer are regularly employed with the companies in question some of them are dispersed throughout the U.S. and I think we run into a severe problem of privacy in addition to the possible harrassment of individuals trying to go evoke this type of information Its cumbersome back and O.K. so don't give me their names I don't want What about the length of exposure tables which length of exposure for living people and period for living people Are those available them summarize of employment Well if you wanted you could have nothing but work history in it It identifiers we would blind out any could give you that a master file that has won't have any individual personal identifiers we That ages need would be fine I don't care to know their names Their would be nice or distribution of their the raw data I can get by with summary ages data I don't A. life table has the number of person we've got all that kind of breakdown the years per age person years group rather than specific flow for each person in the various exposure category so that you could get that indirectly from the life table It would be much less complicated that way These are available to me the life tables Well they certainly will be made available when we have | our general meeting here O.K. Protected Document to Protective order 0148416 JNJ 000042680 -13- JR SS JKW JR SS JR SS JR The point is I've been asked by someone who I have to rely on to pay my salary you know to get this information as quickly as possible and to try to make some head and tail out of it Is it possible that I could come there or you could send me this kind of information about either the living or everybody in terms of these life tables or both Pause ling pause While you are thinking can I ask you a second question about this Wait a second we'll try to work this one out Go ahead shoot your second question There are 405 people that number of people that were from the people you talk about is this the total available or is it a sample It's the total number of people with who worked a minimum of one year with January 1940 1970 and one day of that time had at least one day of that time between 1940 and the current year has to be before of that time had to be at least one day to be after 1940 So as far as you know it's a complete census as opposed to sample It's as a matter of fact probably well it's as complete as we can get considering the problems with the company data and the Health Dept. data what with the Health Dept. data as they had in their data a greater percentage of the people with the greater exposures and the companies had more names but the additional names all seemed to be those people who worked The companies also were who worked there longer much less than a year missing quite a few of periods of time and we like two weeks the people think that this is in some probably due to the sketchy of the companies in the past work histories kept 2 the fact that the companies have changed quite a few of them have changed hands several times and the but we ultimately went with records were lost in the changes the Health Dept. records they're more reliable of their own and there were originally 826 people in the Health Dept. and 405 fit our criteria as <> described I can continue asking questions if you want to continue thinking about the other one We've been trying to get a hold George Lee has been trying to a hold of the death certificates of the people who died and he would like to know if you could provide the information with the state in which the death is registered for If I give you those could you find those regard to three people out..tell me Protected Document to Protective order 0148417 JNJ 000042681 -14- -14- SE SE SE SE SS JR SS JKW SS JR JKW JR JKW JR SS JR Which people One is Treffeley West Shall I continue or wait Treffeley West Who else The second one is Guyett Daniel his third one was Reil the first name is first name Edson and the O.K. I can get that information for you it with me right this second I'll have to our original file I don't look it have up in George asked me to ask a second question with regard to these respiratory deaths he noticed that in your transcript people died you mention that you had excluded people who had influenza and pneumonia and he notes that there from on the list of malignant respiratory deaths who were listed as dying from pneumonia or some Its a coding It's coded by the underlying cause of death Any causes death for the immediate cause they die instantly of pneumonia but that's not the underlying cause of death and that's not the way they're coded by nosologist a qualified nosologist The tables that I sent you where it had the coding that coding by a professional nosologist and there's either three or four columns depends on how many causes we found on them prime cause there's contri- butory case 1 contributory cause 2 contributory cause 3 and the - -. ~ is the cause of death classified under ? As an example congestive heart failure due to due to chronic fibrosis of the lung It would out necessarily as congestive heart failure influenza not be coded It would be coded under the underlying causes That's the ICD column cause of death I think that's it Is that the first The first cause on the death certificate underlying Yes that's what the column ICD contains Yes And then ClC and 3 are .. Contributory cause 1 contributory cause O.K. I understand Protected Document to Protective order 0148418 JNJ 000042682 -15- SS JR JKW SS JR JKW JR So although the immediate cause of death might be pneumonia pneumonia might be a resulting factor from something else and in these death certificates they all were at least according to the physician who attended the patients O.K. Well I'm out of questions now What can we decide about getting this other data I'll tell you George would really like to come out there he's asked me several times he said that I should press that point I don't want to become offensive to you Yes well let me explain my position on this Our general feeling is that we certainly don't intend to shy away and certainly would be more than receptive to interact with other people in the scientific community However we don't wish to be put in a position and we refuse to be put into a position of dealing differentially with management and labano d r my general feeling is that since there are multiple corporations involved there has been multiple requests from multiple corporations would prefer that other than what we're really talking about here that we refrain from any further distribution of hard raw data pending the joint meeting of both and labor and anyconsultants wish that they discuss our to bring to interpretation of very candidly the data the go over and analyses any tendbrought questions that may be resolved of this nature obviously are to brought up Meetings a briefing of our interpretation feeling our interpretation of the data answering any questions and opinions that may be certainly scientifically sounded we have no corner on the market consistent of course but reserving data as we so fit consistent the right to interpret the final with good scientific inter- pretation and in that sense that I'm a little reluctant to openly essentially dump our computer volumes that are least three to four inches thick of them a computer dumpout four Possibly duplicated pages and I means it's hundreds and hundreds of From my personal point of view the problem is that if I was asked to attend such a meeting and I probably will it would be very beneficial to me if I had been able to be look at the data before going to the meeting otherwise it would seem to me that people would have to take what's given at the meeting go back and digest it and then come and ask for a second meeting so our point of view is that we would like to get as much information as possible before the meeting so that we can come prepared to discuss our feelings at that meeting Can I ask you a question briefly Sure Protected Document to Protective order 0148419 JNJ 000042683 JKW JR JKW JR JKW JR JKW JR JKW JR po -16- a If you came to a scientific conference do you leave the scientific conference with the feeling that you had to see somebody's basic raw data and entire distribution before you arrived at a judgement as to the validity of that data and interpretation Forget that you're a consultant for the company Come on When I go to a conference I hear a paper I usually don't understand it so that I have usually to take five minutes what's given after they start I'm in terms of hand outs lost and go home and study it like crazy and look up in the literature what is necessary to understand it and then if I'm interested in pursuing it further get in contact with the people that gave the talk and pursue it The point is I really can't say that I would I wouldn't certainly go to a meeting and call somebody a liar or anything like that Well I'm not asking about liar guilt Scientific protocol I'm just talking about Generally if I go to a meeting other than a statistical kind of conference and people present statistical data my experience that I should question that data I'm not saying you obviously know a lot more than I do about this subject so I'm certainly not questioning you but that's my basic feeling I tend to question data when people do statistics because I've seen so much misuse of it I don't deny that It's too much garbage in garbage out My point still is if I have to come to a meeting whereby I'm going done to talk I would intelligently about important work that's being like to have information before that as opposed to coming in cold and having to say well can we talk in a week or two weeks when I've had a chance to digest the information Well I'll tell you what you tell Mr. Lee that one we will be sending a copy of the paper out at least a week before the meeting o.k. and two at that time with descretion we will provide a couple of pertinent matrixes to give you the answers that you're really after You couldn't send them before that time Well basically we're talking about getting the paper out within essentially a week and two days right now so we are not talking about any lies o.k. The thing is that we've attempted to crank up this meeting to occur as fast as we can however in light of the repetitive requests that have come in from Windsor Minerals Johnson & Johnson & Engelhard I suppose everybody is doing the same thing Protected Document to Protective order 0148420 JNJ 000042684 -17- JKW JR SS JKW JR SS JR If they wish to have the meeting and wish to have the manuscript then you're going to have to sit back and let us finish it o.k. re n. What back will about these three death certificates now can I call or have George call back to get that information or you send it to us We'll send them out We'll send them out we can send them Do you have an address there that I can give you my address it's John Rafter Johnson & Johnson Research Center New Brunswick N.J. 08903. So that's Johnson & Johnson Research Center New Brunswick N.J. 08903 and your phone number Area Code 201-524-5087 Protected Document to Protective order 0148421 JNJ 000042685