STATE OF MINNESOTA AND BLUE CROSS AND BLUE SHIELD OF MINNESOTA,
PLAINTIFFS,
V.
PHILIP MORRIS, INC., ET. AL.,
DEFENDANTS.
TOPIC: TRIAL TRANSCRIPT
TRANSCRIPT OF PROCEEDINGS
DOCKET-NUMBER: C1-94-8565
VENUE: Minnesota District Court, Second Judicial District, Ramsey County.
YEAR: February 27, 1998
A.M. Session
JUDGE: Hon. Judge Kenneth J. Fitzpatrick, Chief Judge
TEXT:
THE CLERK: All rise. Ramsey County District Court is again in session, the Honorable Kenneth J. Fitzpatrick now presiding.
(Jury enters the courtroom.)
THE CLERK: Please be seated.
THE COURT: Good morning.
(Collective "Good morning.")
MR. HAMLIN: Good morning.
(Collective "Good morning.")
TIMOTHY S. WYANT called as a witness, being previously sworn, was examined and testified as follows:
BY MR. HAMLIN:
Q. Good morning, Dr. Wyant.
A. Good morning.
Q. At the close of the day yesterday we were talking about variability of estimate -- of statistical estimates. Could you give the jury some perspective for the testimony that is to come. Could you tell us again what "variability of statistical estimates" means?
A. Well what we're talking about here is the fact that, in part, our estimates rely on national surveys, the National Medical Expenditure Survey and the National Health and Nutrition Survey, and these surveys are not taken of everybody in the United States, they're only taken of samples of 35,000 people, approximately, for the Medical Expenditure Survey, and 40,000 some people for the National Health and Nutrition Survey. So if the federal government had --
-had used the same procedures and gone out and selected a different set of 35,000 people, say, for the National Medical Expenditure Survey, and we had used that different set, then we would have calculated a somewhat different estimate of total smoking-attributable expenditures in Minnesota based on that other possible sample.
Q. Is there a statistical measure of variability that is most appropriate here?
A. Yes, I think so.
Q. And what is that?
A. That's what I would call the relative error.
Q. With the court's permission, could you come down to the flip chart and list the relative errors in the plaintiffs' statistical model.
A. Well the plaintiffs' statistical model is broken into several parts.
Q. Why don't we talk about major smoking-attributable expenditures first.
A. Okay. Major smoking-attributable expenditures is the first part, and we can look at the estimated relative error, and for that category, that number is 41 percent.
Q. What does that percentage mean?
A. What that percentage means is that if the federal government, the Agency for Health Care and Policy Research, had
gone out and repeated the NMES survey a number of times, about two times out of three from those various replications
f that survey, if we applied all of our same methods to calculating smoking- attributable expenditures using those different
surveys, about two times out of three our estimates would fall within plus or minus 41 percent of the actual estimate that
we've presented.
*2 Q. Using that 41 percent, what is the range of dollars from low to high?
A. On that range, if we look at our estimate, which was 558 million, that range goes from 328 million to 788 million.
Q. What is that range called?
A. The name for that range is -- is a confidence interval. That's what that's called in statistics.
Q. What is the most probable estimate of the smoking- attributable expenditures for major smoking-attributable expenditures?
A. The most probable estimate is the estimate that we calculated from the survey that was actually conducted, it's 558 million dollars for major smoking- attributable disease expenditures.
Q. Can you put up the other relative errors for the other parts of the model.
A. Sure.
The estimated relative error for the major diseases is 41 percent; reported poor health -- now we're talking about diminished health status -- 42 percent; or other diminished health, 154 percent; and for nursing homes, 176 percent.
Q. Now if you could flip the page back to the major smoking- attributable diseases, now is the relative -- or excuse me. Is the major smoking- attributable diseases based on the NMES survey?
A. Excuse me?
Q. Yeah. Is the estimate for the major smoking-attributable diseases based on the NMES survey?
A. Well it's based on the NMES survey as well as the claims records and the other data sources we talked about, and the NMES survey that was conducted in 1987 is one of those sources.
Q. Now what NMES sample of the population did you use in calculating the smoking-attributable expenditure of 558 million?
A. The NMES survey we used here was the NMES survey of 35,000 people that was conducted in 1987.
Q. Now flip the chart back to your other list.
Can you tell us what these other relative errors mean.
A. Well all of these relative errors are interpreted in the same way as the relative error for major smoking-attributable diseases for the different groups.
Q. Now let me place underneath the easel Plaintiffs' Exhibit 30203, which has been previously admitted, and that's the state expenditures attributable to smoking by disease category for the full model 1978 to 1996. Do you see that?
A. Yes.
Q. Now let me direct your attention to the list of percentages for estimated relative error on your flip chart.
Now is there anything significant about the pattern of those percentages?
A. Well, yes. One of the reasons for calculating relative errors and one reason they're valuable is to show us in which part of our calculations there's more variability and which parts there's less, and there are two or three things that impact on that. One of the things is: What are you shooting for? In studies like this, if you're looking at only something like a three percent increase in nursing home expenditures --
Q. Perhaps you ought to put that on the easel. Yeah.
A. If we're looking for only a three percent increase, it's more difficult to estimate that with a great deal of precision, and we see that the relative error term is much larger down here. Similarly, when we go up to the other effects of diminished health and the effect we're looking at is only a four percent increase based on our best estimates, the four percent smoking- attributable expenditure, again we have a larger relative error than we see up here.
*3 Now when we get up to reported poor health, we still have a small percentage we're shooting for, five percent when
we're talking about what came
out in our best estimate calculation, but now the epidemiology, another factor, is coming into play. We've taken advantage using the data of the basic notion from Dr. Samet that smoking causes disease which results in health-care costs. In diminished health we have a pervasive kind of disease, according to Dr. Samet, one not readily captured by specific ICD-9 codes, but one that can in part be addressed by reported health status, such as was collected in the National Medical Expenditure Survey, and if we can get additional information like that about disease, we see that the relative error is much smaller than it was for these other groups.
And then when we get up to major disease, we see the same pattern in that the relative error is again much smaller. We're making use of the disease information and the claims data and we're tracking the epidemiology, and we're looking for effects that are bigger based on our estimates of what the smoking- attributable expenditures were.
Q. Now based on your experience, how do these relative errors in the plaintiffs' statistical model compare with relative errors that you've encountered in your work in similar cases?
A. Well in my work in similar cases, estimating total dollars in damage calculations and -- and particularly for -- for
products that cause injury, most of the time no relative error is even calculated in my experience, and most of the time in
fact people can't calculate the relative error because they
don't have surveys of the quality of the National Medical Expenditure Survey that would allow them to do that. They're working from much less reliable sources of information.
The places where typically -- it's very unusual, but where I do see people making something like a relative error estimate in damage calculations, it's usually more of an educated guess. And in particular in -- in the asbestos proceedings in New York, demographers trying to estimate the total number of future claims against Johns-Manville Corporation, they started guessing at -- when they got to about -- they could say about a hundred percent was their calculation based on just some other assumptions. It wasn't a statistical calculation like the ones we're presenting here.
So typically you don't even see these calculations. People calculate these estimates from data which don't even allow this calculation, and educated guesses, when -- on the occasions when an expert is willing to do that, fall right in this range.
Q. Okay. Thank you. You can go back to the stand.
Now if you made the same calculation of error for nursing homes that you did for major smoking-attributable diseases, what would the upper range reach?
A. It would be over 700 million dollars.
Q. Would the lower range reach zero?
A. Yes, it would.
Q. What is the likelihood of the actual smoking-attributable expenditure for nursing homes being zero or less?
*4 A. Based on our work and the medical foundation from Dr. Samet, it's impossible for that to happen. The only way nursing home expenditures, smoking- attributable nursing home expenditures could be negative is if nursing homes were paying the state to take care of people, and I think we know that is not a likely thing to happen. What would have to happen for smoking-attributable expenditures to be zero -- well one thing that would have to happen is that, going back again to the medical foundation from Dr. Samet, smoking causes stroke and strokes can cause conditions that result in nursing home entry if the stroke is severe enough.
Now one thing that would have to be true if there were no smoking- attributable expenditures in nursing homes in
Minnesota during this period is that not one single person over this period from 1978 to 1996, with all the tens of
thousands of people going into nursing homes in Minnesota and being paid for by Medicaid, that not one single time was
that nursing home admission from a stroke caused by smoking that resulted in disabilities sufficient to cause nursing home
entry. That's one thing that would have to be true. That couldn't happen ever once over this entire time period. In addition,
there could be no other condition such as lung cancer causing a period of rehabilitation in the nursing home, that couldn't
happen once either over this
entire 19-year period. And only if those things are true could the smoking- attributable expenditures for the state for nursing homes from 1978 to 1996 be zero. And the numbers are such and the percentages are such that it's just impossible for that to happen.
Q. What is the most likely estimate of the smoking- attributable expenditure in nursing homes?
A. Two hundred sixty million dollars.
Q. Now that's not a guess; is it?
A. No, absolutely not. That's a result of applying our models to the actual claims data and the actual Behavioral Risk Factor Survey data and the actual NHANES survey that was conducted from 1970s through 1992.
Q. And in doing that, you used statistical practices?
A. Yes.
MR. BIERSTEKER: Objection, Your Honor, leading.
THE COURT: It is leading.
Q. Well in doing that, can you tell us what some of the standard statistical practices are that you used?
A. We used standard practices such as the method of attributable risk, the method of maximum likelihood, and those are the methods that were incorporated into the models applied to nursing homes damages.
Q. Now are any of the smoking-attributable expenditures that you discussed
yesterday guesses?
A. No.
Q. Why is that?
A. They're all the results of the application of accepted statistical methods to the best available data for this population.
Q. Now with respect to the most likely estimate of smoking- attributable expenditures in nursing homes, what --
Well let me ask you first: What was that estimate based on in terms of a database of surveys?
A. Well that estimate, in terms of surveys, was based on the one actual NHANES survey that was conducted. We have talked earlier about hypothetical surveys that could be done if you were going to do this over and over again, but of course the federal government hasn't done that. Nobody's done that.
*5 The NHANES survey on which we based the nursing home estimate is the one survey that we have, and we used all the data that's appropriate from that survey.
Q. Now what NHANES sample of the population is the range estimate of zero based on?
A. That's based on a hypothetical survey which has never been taken.
Q. Let's turn now to diminished health other effects. I believe the relative error that you have listed on the chart is 154
percent. Do you see
that?
A. Yes.
Q. Now what would you expect to see in small groups in a category such as this?
A. Well in this category we see a larger relative error, and in that situation you'd expect to see more fluctuation across small groups like the kind we were seeing in the skate board example yesterday.
Q. What is the most likely smoking-attributable expenditure for diminished health other effects?
A. Four hundred seventy-seven million dollars.
Q. Now is diminished health other effects based on the NMES survey?
A. Yes.
Q. What NMES sample of the population did you use in calculating the smoking-attributable expenditure for diminished health mixed effects?
A. The actual sample that was taken of 35,000 individuals in 1987.
Q. Well let me ask you about the other estimates in this range for diminished health other effects, apart from the 477 million. What sample would they be based on?
A. Again, anything other than the 477 million are based on estimates of what would happen if in some hypothetical sense
the survey were redone with a different set of people.
Q. What would it cost to do another NMES sample?
A. Tens of millions of dollars.
Q. Are you aware of any authority on damage estimates that discusses confidence intervals that reach zero?
A. Yes.
Q. Let me direct your attention to Trial Exhibit 24046, a full copy of which is on the ledge next to you, and it's -- you've also got a copy in your demonstrative notebook. No, sorry, in the testimony notebook.
Do you have that in front of you?
A. Yes, I do.
Q. Can you identify that.
A. That's a book called "The Evolving Role of Statistical Assessments as Evidence in the Courts."
Q. And can you identify the editor.
A. The editor is Dr. Stephen Fienberg.
Q. Have you reviewed this particular book in the course of your work in this case?
A. Yes.
Q. And do you find it a reliable authority in the published scientific literature?
A. Yes.
MR. HAMLIN: Your Honor, we offer Trial Exhibit 26046 as a learned treatise under 803(18).
MR. BIERSTEKER: No objection, Your Honor.
THE COURT: Is it 26 or 24? I've got two -- I have two numbers.
MR. HAMLIN: It's 26046.
THE COURT: Court will receive 26046.
BY MR. HAMLIN:
Q. Now Dr. Wyant, could you turn to the page marked "PANEL ON STATISTICAL ESTIMATES AS EVIDENCE IN THE COURTS?" Do you have that?
A. Yes.
Q. And could you --
MR. BIERSTEKER: What page number is that?
MR. HAMLIN: Let me make sure. It's in the first portion of the book. I'm not sure that it has a page number. But if you open it, the first two or three pages, panel members there are listed. Have you got that?
*6 Q. Dr. Wyant, can you tell us what's on that page.
A. This book is a product of a panel. This panel was funded by the National Science Foundation, and the panel is Panel on Statistical Assessments as Evidence in the Courts, and Stephen Fienberg is one of the chairs of the panel, and this panel includes some of the leading names in statistics and economics, including Dr. Samuel Krislov at the University of Minnesota, Dr. James Heckman
at the University of Chicago, and it also includes jurists such as Judge Weinstein in -- in New York.
Q. Where is Dr. Fienberg from?
A. Dr. Fienberg, I believe, is still at Carnegie Mellon University.
Q. And is he in their department of statistics?
A. Yes.
Q. Now what does this book advise about damage estimates which have confidence intervals that reach zero?
A. Well this book says -- which is in agreement with my experience that I described earlier -- that the focus in damage estimates is on calculating the best estimate of damages. It's not unusual that any kind of confidence interval would reach zero, and it really doesn't matter if that happens because the focus, again, is on providing a best estimate, which is exactly what we've done with our models here.
THE COURT: Excuse me.
MR. BIERSTEKER: Your Honor, I would ask if the witness is going to say what the book says, that we be given a page reference instead of just a characterization.
THE COURT: Can we -- can we cite to a specific page?
Q. Can you cite to a page? I believe it's in your excerpt there.
A. Pages 114 to 115.
Q. Now are there additional sources of expenditures of the state of Minnesota and Blue Cross Blue Shield of Minnesota that you did not consider in plaintiffs' statistical model?
A. Yes.
Q. And have you prepared an exhibit of these sources of expenditures?
A. Yes.
Q. Can you turn to Trial Exhibit 30207. Is that the exhibit?
A. Yes, it is.
Q. Can you identify it by title.
A. The title is "The Statistical Model is Lower Than Actual Expenditures."
Q. And was this prepared at your direction?
A. Yes.
MR. HAMLIN: Your Honor, plaintiffs offer Trial Exhibit 30207 for illustrative purposes.
MR. BIERSTEKER: No objection on that basis, Your Honor.
THE COURT: Court will receive 30207 for illustrative purposes.
BY MR. HAMLIN:
Q. Dr. Wyant, I've placed Trial Exhibit 30207 on the easel, and can you describe that exhibit for us.
A. Yes. This lists several categories of expenditures that are not included in the statistical model, they're not addressed by any of the estimates that
I've talked about until now, and these include any estimates for any smoking- attributable expenditures that might be related to secondhand smoke effects. The Environmental Protection Agency, for one example, has issued reports about secondhand smoke and in particular has talked about effects on asthma in children. The entire focus of our statistical model has been on adults, and we have not attempted to calculate any smoking-attributable expenditure estimates for children under the age of 19.
*7 Another category which is not addressed in our model in which there may be smoking-attributable expenditures is the category of increased health- care costs for babies whose mothers smoke during pregnancy. Again, there are estimates in the scientific literature of the total medical costs attributable to smoking due to premature and low-birth-weight babies, but we have not attempted, again, to estimate any smoking-attributable expenditures in our model for anyone under the age of 19.
In our Blue Cross Blue Shield claims data we have looked at rated-fee-for- service plans, but we have not looked at any of their rated HMO plans. We have not looked at expenditures for our government program Minnesota Care.
Q. Let me stop you there, Dr. Wyant. What do you mean by HMO?
A. By that I mean health maintenance organization, managed care plans. That's a different set of Blue Cross business, a
different way they can cover rated groups for an employer or a union.
Q. Let's go on to the next bullet point.
A. There's another state plan, Minnesota Care, the model does not address expenditures for Minnesota Care. And finally, the state expended money for health care coverage of its own employees, and except for a couple of years in which those employees were covered by Blue Cross Blue Shield rated plans, the plans that were in our statistical model, we have not addressed any smoking- attributable expenditures that may have occurred among the state's employees.
Q. Why didn't you address any of these sources?
A. Our focus was on getting the most reliable estimate we could for a reasonable part of the smoking-attributable expenditures of the state and of Blue Cross, and to have gone out and tried to collect additional data and additional information and made additional estimates in these other categories would not have let us make the estimates that we did make to the same degree of reliability.
Q. What if any check of plaintiffs' smoking-attributable expenditures have you done by using sensitivity tests?
A. We spoke of one yesterday involving the National Medical -- excuse me, the National Health and Nutrition Examination Survey in the nursing home estimates where we looked at an age range of 60 to 90 as opposed to 55 to 95.
Q. What if any checks have you done of plaintiffs' smoking- attributable expenditures by comparing them to studies of
disease mortality?
A. Again, we --
MR. BIERSTEKER: Objection, Your Honor, I think we went over this at some length yesterday. Asked and answered.
MR. HAMLIN: Your Honor, this is just a summary, and it lays some foundation for the next question.
THE COURT: All right. I'll allow it just for that purpose.
MR. HAMLIN: Okay.
THE COURT: I don't expect us to go over it again.
A. Yes, we talked about those yesterday.
Q. And have you checked the plaintiffs' smoking-attributable expenditures by comparing them to other studies of health-care costs for diseases caused by smoking?
*8 MR. BIERSTEKER: Same objection, Your Honor.
MR. HAMLIN: And again, Your Honor, we're not going to go over it, it's just a foundational question.
THE COURT: Okay. You can answer it. You may answer.
A. Yes, yesterday we looked at comparisons with the Chrysler study and the Framingham study as two examples.
Q. Have you done any other checking?
A. Yes.
Q. What is that?
A. Well one important check involves a comparison for the major smoking- attributable diseases of the results we get from
the refined model with the results that Professor Zeger described for the core model, the simpler model that also addressed
smoking- attributable expenditures for the major smoking- caused diseases.
Q. Do you have an exhibit setting out that comparison?
A. Yes.
Q. Can you turn to Trial Exhibit 30189. Is that the exhibit that sets out the comparison of the core model results and the partial full model results?
A. Yes.
Q. Okay. Was that prepared at your direction?
A. Yes, it was.
MR. HAMLIN: Your Honor, plaintiffs offer Trial Exhibit 30189 for illustrative purposes.
MR. BIERSTEKER: No objection, Your Honor.
THE COURT: Court will receive 30189 for illustrative purposes.
BY MR. HAMLIN:
Q. Dr. Wyant, I've placed on the easel Trial Exhibit 30189. First of all, can you tell us the title of the exhibit.
A. The title of the exhibit is "Comparison of Core Model to Refined Model, Costs for Major Smoking Attributable
Diseases."
Q. Can you tell us what we see on the exhibit.
A. Again, the smoking-attributable diseases are divided into two categories, the first being lung cancer/COPD, and in the core model described by Professor Zeger there was an estimate of 243 million dollars in smoking- attributable expenditures for lung cancer and COPD. With the refined model that same estimate, that corresponding estimate is 206 million.
The other group of major smoking-attributable diseases is the one we call CHD/stroke that has the other ICD-9 codes in Dr. Samet's list for the major diseases, and the estimate from Professor Zeger's core model is 383 million dollars. And the corresponding estimate for the refined model that I've talked about yesterday and today is 352 million dollars.
Q. Now can you tell us what the total expenditures are for the core model for lung cancer/COPD and the CHD/stroke category which includes other major smoking-attributable diseases.
A. The total for the core model for these two groups of major smoking- caused diseases is 626 million dollars.
Q. And can you tell us the total for the refined model of the lung cancer/COPD category and the CHD/stroke category.
A. Yes. That total is 558 million.
Q. Now is that 558 million a part of the 1.77 billion of plaintiffs' total smoking-attributable expenditures?
A. That's correct.
Q. Prior to reaching your smoking-attributable expenditure - - or that is, the plaintiffs' smoking-attributable expenditure of 1.77 billion, had you reached any other results for the state and Blue Cross Blue Shield of Minnesota?
*9 A. Yes.
Q. What results did you reach?
A. In an analysis completed about June of 1997 we calculated an estimate of 1.42 billion.
Q. And how did you arrive at that estimate?
A. To arrive at that estimate we had had to make a choice between basically two statistical methods, both accepted, one called testimation and one called full model, and at that time we used the testimation method to arrive at the 1.42-billion-dollar figure.
Q. Can you tell us about the testimation method.
A. That involves going into subgroups and making decisions within each subgroup as to what factors to retain in the formulas that are used to calculate the three reductions.
Q. And can you characterize that as conservative?
A. Yes.
MR. BIERSTEKER: Objection, leading.
THE COURT: It is leading.
Q. How would you characterize that approach?
A. One difference between the testimation and the full model approach is that the testimation approach in this situation is likely to be very conservative.
Q. And why? Why is that?
A. Well it's --
The testimation approach does not make use of the law of averages and it does not make efficient use of all of the information across small subgroups, particularly with respect to key factors in this kind of model such as the effect of smoking on different expenditures as we go through the various formulas, and the real effects that you can see, consistent effects of smoking across subgroups, are in some cases discarded by the testimation approach.
Q. Why did you use it?
A. Well both approaches are accepted approaches in statistics, and one of the decision criteria that we used in deciding among approaches was that if there were two approaches that were reasonably similar, we would make a judgment to go with the one that was more conservative.
Q. Well after getting these first results, why did you switch to the full model approach?
A. Well experts for the defendants suggested that using the testimation
approach resulted in some formulas that were arbitrary, and at that time we -- prompted by that, we revisited and relooked at the issue and determined that, indeed, we were not making full use of all the information in the population, we were relying too much on small subgroups. In addition, there were technical reasons involving estimation of relative errors that made the full model approach preferable to the testimation approach. And so we used the full model approach to calculate the 1.77 billion dollars that we've talked about yesterday and today.
Q. Which of these two results is the most reliable?
A. The 1.77 billion dollars.
Q. And that's using the full model approach?
A. That's correct.
Q. Dr. Wyant, have you presented a description of the data, the most important statistical methods, and the main principles underlying the model?
A. Yes.
Q. Have you presented all of the details of the model in your testimony?
A. No.
Q. What --
Can you give us an example of what you didn't tell us?
*10 A. Well, for example, we didn't go over in detail all of the aspects of the bivariate probit maximum likelihood fit for probability of smoking compared to probability of having coronary heart disease, for one example.
THE COURT: For which we're very happy.
(Laughter.)
Q. How long did it take you and your colleagues, Dr. Zeger and Dr. Miller, to build this model?
A. Something over two years.
MR. HAMLIN: Your Honor, I have no further questions of Dr. Wyant at this time.
MR. BIERSTEKER: Your Honor, if we could have a few minutes just to move things around, that would be appreciated.
THE COURT: Why don't we take a short recess now.
THE CLERK: Court stands in recess.
(Recess taken.)
THE CLERK: All rise. Court is again in session.
(Jury enters the courtroom.)
THE CLERK: Please be seated.
MR. BIERSTEKER: Good morning. You may not remember me, I'm Peter Biersteker and I represent R. J. Reynolds Tobacco Company.
BY MR. BIERSTEKER:
Q. Good morning, doctor. It's nice to see you again.
A. Good morning.
Q. Doctor, what you set out to do was to estimate the amount of money that the plaintiffs expended from 1978 to 1996 to purchase smoking-attributable health-care services; right?
A. I believe that's correct.
Q. And you would agree with me, wouldn't you, that smokers might have higher total medical-care expenditures than non-smokers for reasons that have nothing to do with their smoking?
MR. HAMLIN: Objection, it's assuming facts not in evidence.
THE COURT: You can answer it if you know.
A. In some circumstances, sometimes, I assume that that is possible that can happen.
Q. And for that reason, you would agree with me, wouldn't you, that to estimate health-care costs attributable to smoking requires a comparison between average expenditures for smokers and never smokers, all other things equal?
MR. HAMLIN: Objection to form, vague.
THE COURT: You may answer.
A. Well I think a calculation of that sort is needed as long as it's within an appropriate framework, taking into account disease and other factors in the way such as we have done in our statistical model.
Q. You would agree, wouldn't you, that it was a goal in your statistical model to compare like to like?
A. That was certainly one goal.
Q. But we know that all other things aren't the same as between smokers and non-smokers; don't we?
A. Well we suspect that that may be the case, and that's one reason to use other factors in the model so that to a reasonable degree we can be sure that we're comparing like to like.
Q. And in fact we know that smokers and non-smokers differ in many ways that your model doesn't consider; isn't that true?
A. I don't know that that's true after adjusting for the factors that we have in our model.
Q. Doctor, why don't you turn to tab 37 in your book, please.
MR. HAMLIN: Do you have a trial exhibit number?
MR. BIERSTEKER: Yes, that's Exhibit No. BYT000271, already admitted into evidence.
*11 A. I'm sorry, I --
Q. You don't have a copy?
A. I may have, but I'm not quite sure where to look for it here.
Q. Well let's see if I can help you.
A. This?
Q. Yes. It should be at tab 37. I think you've got the wrong volume there, doctor.
A. I'm sorry. So where are we?
Q. This is an article by Dr. Samet; isn't it?
A. This is an --
Are we talking about "Editorial Commentary: New Effects of Active and Passing -- Active and Passive Smoking on Reproduction?"
Q. That's correct.
A. Yes. Yes.
Q. It is written by Dr. Samet; isn't it?
A. Yes.
Q. All right. If you could turn to page 349, please, and I would direct your attention to the second full sentence on that page. Do you see that? Do you see it, doctor?
A. Yes.
Q. And doesn't Dr. Samet say, "As the prevalence of smoking has declined, smokers have become increasingly distinct from nonsmokers in other aspects of life-style as well. Smokers today tend to have less education and lower income, to drink more alcohol, and to follow a less healthy life-style than nonsmokers."
Did I read that correctly, sir?
A. You read that correctly.
Q. And didn't the Surgeon General also in his 1989 report indicate that there were differences between smokers and non- smokers?
MR. HAMLIN: Objection, Your Honor. Could we have a page number so we can know what he's talking about?
THE COURT: Can you cite us?
MR. BIERSTEKER: Absolutely.
BY MR. BIERSTEKER:
Q. Doctor, why don't you turn to the 1989 Surgeon General's report, Plaintiffs' Exhibit 3821, already in evidence, and let's go to page 139. Have you found it yet?
A. Yes.
Q. And there's a column on that page marked "Other Changes in the Cigarette Smoking Population." Do you see that?
A. Yes.
Q. And doesn't the Surgeon General say, in the second sentence of the second paragraph under that heading, "There is some evidence that socioeconomic difference -- differentials in smoking rates have widened." Do you see that?
A. Yes. It's -- it's reasons like this that are precisely behind the inclusion of other factors in our refined model.
Q. I see. And doesn't he say that the proportionate decline in adult smoking rates between 1965 and 1985 was the highest for people who graduated from college and the lowest for those who had not completed high school?
A. Yes.
Q. And he goes on to say that there's been a bigger decline in smoking among white collar workers than among blue collar workers; right?
A. Yes, that's correct.
Q. And then if you skip down to the last paragraph on this page, the Surgeon General says, "Cigarette smokers have higher rates of alcohol use, are more sedentary, and are less likely to wear their seatbelts." Do you see that?
A. Yes.
Q. Now doctor, you don't know in your data set whether all things are equal between smokers and non-smokers; do you?
MR. HAMLIN: Objection, vague.
THE COURT: You may answer that.
*12 A. I'm sorry. We don't know in our what?
Q. In your own data set whether all other things are equal as between smokers and non-smokers, other than their smoking.
A. No. What we know is we have applied reasonable and accepted methods to an extent that any statistician would say -- and certainly the statisticians in our group -- that to a reasonable degree we've compared smokers and non-smokers on a like- to-like basis.
Q. Well let's ask about some of that. When you were building your model, did you look in your data set to see whether smokers and non-smokers differed with respect to the rates at which they had accidents?
A. No.
Q. And if smokers have more accidents and spend more money on health care because -- to treat injuries caused by those accidents, then those dollars will not be removed by your third reduction step; will they?
A. The third reduction will -- will allocate dollars to smoking only to the extent, in all disease categories, that smokers on average compared to similar non-smokers have higher expenditures. And the way it works is in the other direction as well, to the extent there are higher expenditures for non-smokers than smokers, those are taken out.
Q. Okay. So if smokers have more accidents and spend more money on health care to treat injuries caused by those accidents, your model is going to attribute those dollars, those extra dollars, to smoking; right?
A. No. It's going to make the calculation that -- that I just said. It's not going to go in and make specific attributions for specific injuries or diseases. It makes an overall calculation comparing smokers to non-smokers.
Q. If smokers and non-smokers had exactly the same costs for everything else other than car accidents, your model -- and smokers had more car accidents and more health-care costs for that, your model would attribute those dollars to smoking; wouldn't it?
MR. HAMLIN: Objection, foundation, assuming facts not in evidence.
THE COURT: No, you may answer that.
A. Well I mean that's postulating a circumstance that doesn't -- in my experience with these data isn't remotely close to what we see there.
Q. Doctor, I'm just trying to understand how the calculation works. Can you answer the question, please.
MR. HAMLIN: Your Honor --
THE COURT: Just a moment, please. Do not instruct the witness, please.
MR. HAMLIN: I would also ask that the witness be allowed to answer the question fully.
THE COURT: And allow him to finish his answer.
MR. BIERSTEKER: Were you finished, doctor?
THE WITNESS: No. I'm sorry, I'm lost as to where we are.
Q. If smokers and non-smokers had identical medical expenditures for everything except automobile accidents, and the smokers had higher costs to treat injuries from automobile accidents, isn't it true that your model would attribute those higher costs to smoking?
A. But they don't have identical costs for everything. And one thing that you see is that they have, on average, higher costs for smoking-related conditions, which, if we're talking about auto accidents, if my understanding of Dr. Samet is correct, for example, it can cause complications of treatment in those auto accidents. That's why the importance here is comparing smokers to non-smokers on an average basis within a structure of a disease model as we've done, making reasonable attempts to compare likes to like.
*13 Q. Can you answer the question I asked, doctor?
MR. HAMLIN: Objection, Your Honor.
THE COURT: It's argumentative.
MR. HAMLIN: I'm sorry, it was asked --
Q. Let's say that smokers cost on average a hundred dollars more than non- smokers because of lung cancer, but let's say smokers also have fifty dollars more costs because they have more automobile accidents. Does your model attribute one hundred dollars to smoking, or 150?
A. That's such a simple excerpt, I -- I don't even know how to characterize it in terms of the model. I --
The model is what it is, and it is what I've described.
Q. If smokers have more sprained knees, I think was the example you gave yesterday, than non-smokers do, your third reduction step will not remove the extra costs for sprained knees that smokers have; right?
A. The only way that -- that sprained knees can be included as part of smoking-attributable expenditures is if, in the context of comparing smokers to non-smokers, all medical conditions, that smokers tend to have higher costs compared to similar non-smokers for sprained knees.
Q. And so if smokers have higher costs than similar non- smokers for automobile accidents, those costs will be attributed to smoking in your models; right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: It's been asked and answered.
Q. Did you compare smokers and non-smokers in your data set to determine whether they differed with respect to exercise levels?
A. We made some calculations using the Behavioral Risk Factor Survey as to how exercise affected the relationship of smoking to health-care utilization.
Q. In the calculations that you did to generate your estimates, you didn't include exercise as a variable; did you?
A. Exercise was not one of the explicit factors in the refined model.
Q. Do you know what happens if exercise is added to the refined model, say for coronary heart disease and stroke?
A. I have no accurate way of doing that. And in general in these studies, I would think that's a dangerous thing to do because of the kind of synergies and over-adjustments that Dr. Samet mentioned.
Q. Did Dr. Samet tell you not to put exercise in your CHD/stroke model?
A. No, he didn't. We had many general discussions with him on the nature of these diseases.
Q. Couldn't you determine what happens in your CHD/stroke model if you add exercise just by putting it in?
A. I wouldn't know how to interpret that, so I don't know that you could tell what happens in terms of identifying some change in smoking-attributable expenditures. One of the problems is that when people get sick, they stop exercising, and what you're doing is simply taking the sickness out of the model if you put exercise in and don't do it with a great deal of care.
Q. Did you include depression in any of your models?
A. No.
Q. Do you know if smokers are more depressed than non- smokers?
A. I've seen some articles, some of which show that smoking precedes depression, some of which discusses relationships of one kind or another, but that's about the extent of my knowledge, other than that as Dr. Samet, I believe, testified to depression being a consequence of having something like lung cancer.
*14 Q. You can be depressed for other reasons, though, too; right?
A. Sure.
Q. So your model doesn't take into account that depression; right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: It's been asked and answered.
Q. Do you know what happens if depression and exercise together are added to your CHD/stroke model?
A. Well again I wouldn't know how to even do that in any statistically meaningful way, and -- and so one result of that is I don't know what would happen with some theoretical addition of that.
Q. Well I don't understand. How could you add being overweight to your model but not know how to add exercise and depression?
A. The data are in different forms and there are different considerations in terms of medical foundation and what are in similar articles.
Q. There's no medical foundation that you know of linking exercise to health status or to health-care expenditures?
MR. HAMLIN: Objection, mischaracterizes his testimony.
THE COURT: No, you may answer.
A. No, that's not what I said.
Q. Okay. So there is a medical foundation, then, for including exercise; right?
A. No, I'm not saying there's a foundation for including exercise. Exercise may be clinically relevant, certainly, to certain conditions. That doesn't mean that including exercise in a statistical model for a population of the effect of smoking on expenditures is a reasonable thing to do.
Q. So it's not reasonable to include exercise, but it is reasonable to include seatbelt use?
A. That was our decision.
Q. And Dr. Samet didn't tell you to do that; did he?
A. No, he made no specific recommendations on individual factors of that sort.
Q. Did you compare smokers and non-smokers in your data set with respect to diet?
A. Not in our data set, no.
Q. Did you compare smokers and non-smokers in your data set with respect to injuries, the rates at which they had injuries?
A. No.
Q. Did you compare smokers and non-smokers in your data set to see which group had more mental health problems?
A. We made a calculation at one point in a sensitivity test of some mental conditions on what would happen with the model results.
Q. And what happened?
A. There was no appreciable effect.
Q. Were those results produced to us?
A. The programs would have been if we had them.
Q. Did you compare smokers and non-smokers to determine whether or not smokers were more likely to suffer from conditions caused by drug abuse?
A. Ah, we didn't make that comparison, no.
Q. Did you compare smokers and non-smokers to determine whether or not if smokers drank more heavily and were more likely to be alcoholics?
A. No.
Q. Did you compare smokers and non-smokers with respect to family history of disease in your data set?
A. In this data set?
Q. Yes.
A. No.
Q. Did you compare smokers and non-smokers with respect to occupation?
A. No.
Q. Did you compare smokers and non-smokers with respect to environmental exposures in your data set?
*15 A. No.
Q. Do you know whether smokers or non-smokers in your data set have more diabetes?
A. As a whole, no.
Q. Do you know whether smokers or non-smokers in your data set have more social support?
A. No.
Q. Do you know whether social support is an important factor in determining whether somebody goes into a nursing home?
A. Social support is one factor, it's my understanding, of going into a nursing home.
Q. Did you compare levels of stress among smokers and non- smokers in your data set?
A. No.
Q. Did you determine whether or not smokers or non-smokers were more likely to be disabled in the public aid population?
A. I wouldn't know what the relevance of such a calculation would be.
Q. Don't people who are disabled and on public aid cost a lot more money than people who aren't?
MR. HAMLIN: Objection, assuming facts not in evidence. Counsel is testifying.
THE COURT: You may answer that, if you know.
A. Well I don't know that that's true, but I also know that one way to achieve disability status on Medicaid is through smoking-attributable diseases such as lung cancer.
Q. But you don't know whether disability is a very expensive category within the Medicaid program; is that right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: It's been asked and answered.
Q. Now your nursing home model estimates that the state spent about 260 million dollars for smokers who entered the nursing home because they had a smoking-related disease or condition; is that right?
A. That's correct.
Q. And your model compares time spent in nursing homes by never and ever smokers who are of the same age and the same sex; right?
A. That's correct.
Q. And I'd like to explore for illustrative purposes a little bit how you did this calculation by talking about the 94- year-old women.
So one of the age and gender groups that you looked at was to compare ever and never smokers, the amount of time they spent in the nursing home, who were women who were age 94; right?
A. That was one part of the calculations.
Q. And isn't it true that there were two 94-year-old women ever smokers in the nursing home in your national data set?
A. I believe that's correct.
Q. How many non-smoking 94-year-old women were in the nursing home, do you know?
A. I don't recall.
Q. And all that your nursing home model considers for these 94-year-old women are the time they spent in the nursing home and whether they ever smoked; right?
A. And age.
Q. And age.
A. Right.
Q. They're all 94-year-old women.
A. Okay.
Q. And for the 94-year-old women, the ever smokers spent more time in the nursing home than non-smokers did; right?
A. Correct.
Q. And you apply that difference and the time spent in the nursing home for those two 94-year-old women compared to however many non-smoking 94-year-old women there were in your sample, you apply that difference to every woman who is 94 years old in the state of Minnesota from 1978 to 1996 who spent time in a nursing home who ever smoked; right?
*16 A. That's the calculation that's included in the model, yes.
Q. And you did the same thing for 93-year-old women, and for 92-year-old women and for 91-year-old women and 90, all the way down to age 55; right?
A. That's correct.
Q. And you did the same thing for men for that same age span, one unit at a time; right?
A. That's right.
Q. Now someone was an ever smoker if they smoked a hundred cigarettes or more in their lifetime; correct?
A. That's correct.
Q. And it didn't matter when people here in Minnesota smoked to your model; did it?
A. The model took that definition, which is a standard definition, and calculated using that definition, so that you got appropriate differentials according to that definition of smokers versus never smokers.
Q. Did it matter what kind of cigarettes the people smoked, whether they were low tar or high tar?
A. Whatever they smoked, it was included in the average calculations.
Q. And it doesn't matter whether the person quit or when they quit; right?
A. Well I don't know if it matters or not. To the extent that it matters, it's going to affect the averages we used, and the averages will reflect exactly what the smoking patterns were in the groups in that national population.
Q. Does it matter for the people here in Minnesota?
A. Does what matter?
Q. Does it matter for the people here in Minnesota whether they ever quit or when they quit?
A. In what sense?
Q. Well, you put women into your model who were here in Minnesota who you estimate were smokers; right?
A. Yes.
Q. Okay. For those women here in Minnesota, not in your national data set but here in Minnesota, does it matter when they quit or if they quit smoking?
A. I'm not sure of the point of the question. A reasonable way to estimate the effects of smoking on use of nursing homes in Minnesota is to take a standard definition of smoking versus non- smoking and apply that in the way we did through our calculations stratified by age and sex.
Q. Is that a yes?
A. I'm not sure --
MR. HAMLIN: Object.
A. -- if I really understand the question.
Q. Well let me try one more time then.
Does it matter to your calculation whether anybody here in Minnesota, in your Minnesota data set, quit smoking or when they did, --
A. Well I don't know --
Q. -- for purposes of your nursing home calculation?
A. We didn't take that into account in the Minnesota data because it wasn't available and because what we were doing made for a reasonable application of statistical methods to answer the question we were addressing.
Q. It doesn't matter to your model, does it, whether a person entered the nursing home with lung cancer or Alzheimer's disease; does it?
A. It matters to the model to the extent that whatever those reasons are, they're picked up in our calculations in a way that reflects the average tendency of those things to happen in the population.
Q. You compared smokers, people who had smoked a hundred cigarettes a day -- or excuse me, a hundred cigarettes or more in their lifetime, to non- smokers, and you looked at the difference in time that they spent in the nursing home, and it didn't matter why the person went into the nursing home; right?
*17 MR. HAMLIN: Objection, asked and answered.
THE COURT: No, you may answer that.
A. We didn't make any attempt to try and figure out exactly why for each individual they might have gone into the nursing home. We looked at the population and calculated statistics about the population of people staying in nursing homes or not staying in nursing homes.
Q. Well let me ask this question: Does your model differentiate between those smokers in your sample who entered the nursing home because they had a disease or condition related to smoking, and those smokers who entered the nursing home with some other kind of condition?
A. It doesn't try to differentiate smokers beyond what I've already said.
Q. So it doesn't do that; is that right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: I believe it's been answered.
Q. Your national sample, however, contains a lot more information about these women and about all of the people in your national sample who entered the nursing home than your model actually used; right?
A. It does contain more information.
Q. And we can illustrate some of the additional kinds of information by looking at the 94-year-old women.
Have you looked at the 94-year-old women, the two smokers who went into the nursing home in your national sample?
A. I've seen some printouts of that information.
Q. Do you know what the diagnoses were for those two women?
A. No.
Q. Do you know when they started to smoke?
A. I don't recall.
Q. Do you know how many cigarettes they smoked?
A. I don't recall that either.
Q. Do you know whether they had any conditions at all that Dr. Samet or anybody else has said are related to smoking?
A. Dr. Samet, in -- in my understanding, has said that a variety of conditions could be complicated by smoking as well as explicit smoking-caused conditions in his charts.
Q. Did Dr. Samet tell you that depression was caused by smoking?
A. As a consequence of lung cancer.
Q. Okay. Did Dr. Samet tell you that paranoia is caused by smoking?
A. No.
Q. Did Dr. Samet tell you that non-psychotic mental disorders are caused by smoking?
A. No.
Q. Did Dr. Samet tell you that depression related to your -- your spouse's death is caused by smoking?
A. He didn't tell me that it was, he didn't tell me that it wasn't.
Q. Do you think it is?
A. I don't know.
Q. Do you think a history of manic depression is caused by smoking?
A. I am not a medical expert.
Q. Did Dr. Samet tell you that manic depression was caused by smoking?
A. He did not tell me that.
Q. Did Dr. Samet tell you that history of nervous breakdowns is caused by smoking?
A. No.
Q. Did Dr. Samet tell you that any specified psychoses are caused by smoking?
A. No.
Q. Anxiety, did Dr. Samet tell you that anxiety is caused by smoking?
A. Only to the extent that, again, the treatment of almost any of the wide variety of conditions could be complicated by the fact of a smoking-related illness.
*18 Q. Do you know whether or not either of the two 94-year-old women in your national sample who were in the nursing home and who were ever smokers were in the nursing room -- nursing home for anything that had anything to do with their prior smoking?
A. In any statistical study like this you never know that about any individual. The whole point of a study like this is to take individuals in a population and make appropriate calculations to take into advantage the law of averages.
Q. And your model doesn't take into account the diagnoses of anybody in your national sample who went into the nursing home; right?
A. Not the model that we've presented here today.
Q. Now your model does estimate and attribute to something 87 million dollars in nursing home expenses for 94-year-old women; doesn't it?
A. No. I think that's a misleading way to characterize it. I've not used the model to make estimates for groups such as 94- year-old women. We used the model in totality to estimate a population total and allow for fluctuations from age group to age group.
Q. And the way you got your population total was to take an estimate for each age and gender group and add them up; right?
A. I don't know as I would characterize, for each age and gender group, the number as an estimate. What you have for each age and gender group is a contribution to the overall total. It is not the estimate that I would make if I had the task of estimating for a particular age or sex group.
Q. Let me ask the question a different way.
If we recalculated your model as a sensitivity test the way you testified about yesterday, and this time we just excluded 94- year-old women, wouldn't your estimate be 87 million dollars lower?
A. I think that would be an unreasonable thing to do. A balanced way to do a sensitivity test would be to apply a rule that says something like the one we used. Let's take two age groups, look at where 95 percent of the data are, and let's do a sensitivity test where we take out people that could fluctuate in either direction.
Q. Doctor, I understand you think it might not be a reasonable thing to do, but I want to know what the result would be if you did it.
MR. HAMLIN: Objection, asked and answered, and also object to counsel's characterization of the statement.
THE COURT: Yes, counsel, do not characterize the statement. Just ask a question, please.
MR. BIERSTEKER: All right. Well Your Honor, I move to strike the last answer as not responsive to the question, and I'd like an answer to the question.
THE COURT: Okay. Re-ask the question, counsel.
BY MR. BIERSTEKER:
Q. Doctor, if you recalculated your nursing home model and we did a sort of sensitivity test like you talked about yesterday, and for purposes of my sensitivity test I just want to exclude 94- year-old women, the result I get will be 87 million dollars lower than the estimate that you presented; won't it?
MR. HAMLIN: Objection, asked and answered.
*19 THE COURT: It's been asked and answered.
Q. Doctor, the result from your nursing home model based on all the people in your national survey, not just the 94-year-old women, now those results don't mean that smoking actually causes increased nursing home expenditures; do they?
A. We are not trying to establish cause with our model. We're assuming cause based on the medical foundation of Dr. Samet from such factors as smoking causes stroke, stroke can cause nursing home entry, and other conditions in the same manner. All we're trying to do with the model is measure the extent to which in this population that causation is reflected in increased medical expenditures.
Q. Okay. Do you assume, then, that the results of your nursing home model based upon your national survey are a reasonable estimate of the amount of nursing home usage actually caused by smoking among people on Medicaid here in Minnesota?
MR. HAMLIN: Objection, Your Honor, that's asked and answered.
THE COURT: Well, you can answer it again.
A. I think what the model gives is a reasonable measure of the extent of those dollars.
Q. And you make that causation assumption, a reasonable estimate of increased nursing home usage actually caused by smoking, even though your model doesn't consider the people in your national sample, what diseases they had; right?
MR. HAMLIN: Objection, mischaracterizes the testimony.
THE COURT: You may answer that.
A. I'm sorry.
Q. You make that causation assumption that your estimate is a reasonable estimate of the amount of nursing home usage actually caused by smoking even though your model doesn't take into account what diseases a person had; right?
A. The model is a simple model in that it compares smokers to non-smokers, taking into account their age and sex.
Q. And not their disease; right?
A. That is not taken into account.
Q. And you make that assumption even though your model doesn't establish a causal relationship between smoking and nursing home disease; correct?
A. We assume, going into the model, that there is causation, and all we're trying to do is measure the extent to which that causation results in smoking- attributable expenditures.
Q. Doctor, does your model establish a causal relationship between smoking and nursing home usage?
MR. HAMLIN: Objection, asked and answered.
THE COURT: I think it's been asked and answered now.
BY MR. BIERSTEKER:
Q. Doctor, at the time you made your causation assumption, you didn't have any information that established a causal relationship between smoking and nursing home usage in the epidemiology; did you?
MR. HAMLIN: Objection, foundation. Counsel is testifying.
THE COURT: You may answer that.
A. I believe that my causation assumption came from Dr. Samet in the ways I've described, and at what point in time we had that discussion I couldn't tell you.
Q. Doctor, why don't you turn to volume three of your deposition at page 389, and I direct you to line five, please.
*20 A. 389?
Q. 389, yes, sir.
Are you with me?
A. Yes.
MR. BIERSTEKER: 389, line five.
Q. I asked this question:
"Doctor, my question was: Is there any information that establishes a causal relationship between smoking and nursing home usage?"
And then there's an objection and a response. And then your answer on line 12:
"I don't know of any."
Is that the question you asked and the answer you gave in your deposition, sir?
A. Well just before that I state very clearly that it's my understanding there are causal relationships between smoking and diseases that would lead to nursing home usage.
Q. And it also states very clearly here that you had no information, regardless of what your understanding was, that smoking causes -- there's a causal relationship between smoking and nursing home usage; right?
MR. HAMLIN: Objection, Your Honor, that's a misuse of the deposition. Dr. Wyant answered the question fairly and directly and he pointed out to counsel exactly how he answered the question, and now counsel is misconstruing the deposition and using it inappropriately.
THE COURT: You may answer the question.
A. I'm sorry, now what is the question?
Q. Doctor, the question I asked was: Is there any information that establishes a causal relationship between smoking and nursing home usage, and you answered I don't know of any; is that right?
A. That's correct.
Q. Now on your testimony yesterday, I believe you referred to an article, Plaintiffs' Exhibit 18943. It's at tab four in your book. Would you turn to it, please.
Do you have it?
A. Yes.
Q. All right. Now this article reviews in the second paragraph on the first page, sir, some of the previous research on risk factors for going into a nursing home; doesn't it?
A. Yes.
Q. And the factors it identifies in that paragraph are advanced age; correct?
A. Yes.
Q. Being caucasian or white?
A. Yes.
Q. Having a physical disability; correct?
A. Yes.
Q. Having a mental impairment; right?
A. Yes.
Q. Living without a spouse; correct?
A. Yes.
Q. And your model takes into account none of those factors other than age; right?
A. Again, as I've stated, it takes them into account to the extent that these are in the averages for the populations that we've compared.
Q. And this study concluded in part -- if you turn to page 638, sir, the second-to-last line in the first column, that "smoking history was not a risk factor in nursing home usage by women."
A. That's exactly consistent with what I talked about yesterday. These are smoking effects on top of the effect of smoking on any conditions such as neurological conditions and stroke, and that there is an increased risk identified here for men but not for women.
Q. And the authors go on at page -- the next page, 639, at the top of the page in the discussion section and they say, "These findings, in general" -- talking about the results of their research -- "were similar to previous research on the risk of nursing home admission and illustrated that advanced age, restricted outside mobility, basic ADL disabilities" --
*21 "ADL" means Activities of Daily Living; correct?
A. That's correct.
Q. -- "were the strongest individual predictors of entering a nursing home during a 10-year period." Right?
A. This is a result that's entirely consistent with our estimates of only three percent smoking-attributable nursing home expenditures.
Q. And doctor, none of these criteria other than age, again, was taken into account in your model; right?
A. Not directly.
MR. HAMLIN: Objection.
Q. Did you look at your data set to see whether smokers or non-smokers had more restricted outside mobility?
A. No.
Q. Did you look at your data set to see whether smokers or non-smokers had more basic activities of daily living disabilities?
A. No.
Q. Now you mentioned, I think, a little while ago in your testimony, other estimates with respect to nursing homes, and in fact you made some other estimates of smoking-attributable expenditures for nursing homes; didn't you?
A. We have made some other estimates of smoking-attributable expenditures for nursing homes, yes.
Q. And Dr. Miller in fact had principal responsibility for making the nursing home calculations, at least up through June of last year; right?
A. Well it was a collaborative effort, and we all worked on it to one degree or another.
Q. Well Dr. Miller's answers with respect to the refined --
First let me ask this: The refined model includes the nursing home analysis; is that correct?
A. That's right.
Q. And Dr. Miller's answers about work on the nursing home refined model, at least up through June of 1997, are authoritative answers; aren't they?
A. I'd have to see the questions and the answers.
Q. Well why don't we look at your deposition, sir, volume one, page 148.
A. Of my deposition?
Q. Yes, your deposition, sir. And I'll refer you to line 15, and I asked this question --
There's a series of two questions and two answers.
"Well let me ask you this." And I was directing this my question to you, Dr. Wyant. "Dr. Miller was the person who concentrated on the refined model; right?
"Answer: Yes.
"Question: Are his answers to question, then -- to questions, then, about that model the authoritative answers?
"Answer: I would presume so."
Does that refresh -- refresh your recollection, sir?
A. Yes.
Q. Now Dr. Miller said that your other estimates of smoking- attributable expenditures for nursing homes that he calculated before June were all over the map; didn't he?
MR. HAMLIN: Objection, Your Honor, it's improper use of a deposition of another witness with Dr. Wyant.
THE COURT: Sustained.
BY MR. BIERSTEKER:
Q. Do you know whether or not the other estimates of nursing home expenditures that were made prior to June of 1997 were all over the map?
A. I wouldn't characterize them that way.
Q. Do you know if Dr. Miller characterized them that way?
MR. HAMLIN: Objection, Your Honor, the same objection.
*22 THE COURT: The objection is sustained.
Q. Do you remember whether any of those other estimates of smoking- attributable expenditures for nursing homes showed that smokers didn't cost the state of Minnesota any money?
A. I don't recall.
Q. And we'll never know; will we?
A. What do you mean?
Q. Weren't the results from those analyses thrown away?
A. Well when we started analyzing the nursing home data, it was a difficult data set. There were a number of models that gave very high estimates, as I recall, but there were other estimates as well, but since none of them we considered reasonable or correct, we went on to calculating a more reliable model, which is the one that we presented here today.
Q. Do you know if those results were thrown away or not?
MR. HAMLIN: Objection, Your Honor, it's irrelevant, asked and answered.
THE COURT: You may answer.
A. I don't have any such results around.
Q. And if you don't have them, where would they be?
MR. HAMLIN: Same objection, Your Honor, it's irrelevant.
THE COURT: You may answer that.
A. I have no knowledge of any such results.
Q. Let's talk about your diminished health status model. Now none of the people in your diminished health status model were treated for any smoking- related disease in the entire year; right?
A. Well I wouldn't characterize them that way.
Q. Well they didn't -- they didn't have any treatment for lung cancer, for example; right?
A. Well they were defined as not having treatment for one of what we've called the major smoking-attributable diseases.
Q. Oh, I forgot the word "major." All right. Let me ask the question again.
The people in your diminished health status model were not treated for any of the major smoking-attributable diseases; is that right?
A. I believe that's correct.
Q. And your total estimate for those people, putting together the two diminished health estimates, is 952 million dollars; right?
A. I haven't done that calculation, but it sounds approximately right.
Can I check it?
Q. Sure. Go ahead if you want.
A. I'm sorry, what was your figure?
Q. I thought it was 952 million dollars.
A. That looks approximately right.
Q. Now the individuals in this diminished health status model could have had any medical condition, injuries from a car accident, poisonings, mental illness, AIDS, drug abuse, ingrown toenails; right?
A. Any -- any individual could have any array of injuries, diseases.
Q. And you base your estimate for diminished health status on how good these people say they feel; right?
A. We use a standard method for these kinds of studies, as I've said, looking at self-reported health status.
Q. Your answer -- I'm sorry. Did I cut you off? I didn't mean to.
A. I believe I've finished.
Q. Were you finished? Okay.
What you used was the answer to the question: "In general, would you say that your health is excellent, good, fair or poor?" Right?
*23 A. Yes.
Q. And that's a subjective measure of health status; right?
A. Yes.
Q. No doctor ever examined these people and ranked their health status according to some objective criteria as excellent, good, fair or poor; right?
A. Well I don't know that one did or didn't, but the data we were using is a self-assessment on health status.
Q. And you compared the reports of current and former and never smokers by how good they felt, how they answered that question; right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: You may answer that.
A. That part of the calculation involved looking at the extent to which smokers and never smokers responded to that question.
Q. And you referred yesterday generally to some literature about self- reported health status; didn't you?
A. I may have. I don't recall.
Q. Well are you aware of any article in the published literature that compares the self-reported health status of smokers and non-smokers who don't have any major smoking-related disease?
A. No, I don't think anyone in the studies that I've seen have had access to something like our claims data where we could separate out the model by people who have major smoking- attributable diseases and people who don't.
Q. Are you aware of any literature that compares the self- reported health status of people who are on Medicaid?
A. I don't recall any.
Q. Isn't it true that there's a relationship reported in the literature between self-reported health status and exercise?
MR. HAMLIN: Objection, Your Honor, foundation, counsel is testifying, assuming facts not in evidence.
THE COURT: Sustained.
Q. Well doctor, why don't you turn to tab 13 in your book, which would be Plaintiffs' Exhibit 26041.
Doctor, is this one of the articles which you read in preparation for your testimony?
A. Yes.
Q. In fact, it was an article that the plaintiffs designated; isn't it?
A. I believe so.
Q. And if you'll turn, then, sir, --
Well let me ask you this: Is it authoritative and reliable?
A. I believe so.
MR. BIERSTEKER: Okay. Your Honor, I would move the admission of Plaintiffs' Exhibit 26041 as a learned treatise.
MR. HAMLIN: No objection, Your Honor.
THE COURT: Court will receive 26041.
BY MR. BIERSTEKER:
Q. And doctor, if you'll turn to page 34 of this exhibit, there's a figure, Fig. 2, and it shows a relationship between exercise and good self-reported health status; doesn't it?
A. Yes.
Q. And if you'll look at Fig. 3 on the same page, it reports a relationship between good self-reported health status and how many hours of sleep you get a night; doesn't it?
A. I think that's true. And all of these follow the leading charts and conclusions in here that look at odds of reporting poor health declining with smoking.
Q. Ah, yes. But these were all people who smoked, not just - - not just people who didn't -- not smokers who had no major smoking-related disease; right?
*24 A. But the incidence of those diseases is such that I reasonably conclude that most of these people did not have major smoking-attributable diseases.
Q. Doctor, people with heart disease who smoke are in this; right?
A. It's possible.
Q. Well do you know or not?
A. No.
Q. The article also reports a relationship between self- reported health status and alcohol consumption; doesn't it? I think if you'll turn to Fig. 6 on the next page.
A. It reports a relationship.
Q. And doesn't it also report a relationship between self- reported health status and weight among both men and women in Figs. 4 and 5?
A. Yes.
Q. And your model of predicting self-reported health status does not include exercise, sleep, alcohol consumption; does it?
A. But it includes many other factors that are not listed here, and some that are listed here.
Q. Yes. You include weight; don't you?
A. Yes.
Q. And you include smoking.
A. Yes.
Q. Now does this model consider any differences in the accident rates of smokers and non-smokers in your data?
MR. HAMLIN: Objection, vague. We're -- I think we're unclear as to whether counsel is talking about this article or the statistical model.
THE COURT: Do you understand the question?
THE WITNESS: Could you ask it again?
MR. BIERSTEKER: Sure, I'd be happy to.
Q. Put aside the article. Does your model for self-reported health status take into account differences, if any, in the rates at which smokers and non- smokers have accidents?
A. As I've said before, the model takes into account all conditions and diseases by the same method, by looking at the extent to which smokers and similar non-smokers have higher or lesser expenditures after structuring our model along the lines of disease as I've laid out.
Q. And you don't know whether those similar smokers and non- smokers have the same rate of accidents; right?
MR. HAMLIN: Objection, Your Honor, asked and answered. This has been asked several times and answered several times.
THE COURT: It's a little different question. You can answer it.
A. I'm sorry?
Q. And you don't know whether the smokers and non-smokers who you are describing as similar have different rates of accidents -- of accidents; do you?
A. Based on our discussions, our reviews of the literature, we think that variables and factors in our model to a reasonable degree made smokers and non- smokers similar for purposes of the estimation we were making.
Q. But you haven't looked in your data set to determine whether or not there are differences between smokers and non- smokers, with respect to accidents and with respect to poisonings and with respect to exercise and with respect to drinking and with respect to dietary habits and with respect to mental state, to determine whether or not those differences are adjusted for. They're just not taken into account; right?
MR. HAMLIN: Objection to form, and asked and answered.
THE COURT: You may answer.
*25 A. Those are all taken into account in the way I've already stated, which is a reasonable way to approach it, and one entirely consistent with the way these studies are done in the literature.
Q. All right. So you say you've taken them into account. Is there -- is there --
You had that list of factors up on the monitor yesterday that you considered in your refined model. Was accidents among those?
A. No.
Q. Okay. Were poisonings among those?
A. There would be no reason. It would be improper to put those in as a factor in the model because of the fact, as Dr. Samet says, treatment for any of those could be complicated by having a smoking- related condition.
Q. Let's go back to a hypothetical similar to one I'd asked you earlier. If we're comparing the self-reported health status of smokers and non-smokers and they're alike in every way except smokers have more accidents, isn't it true that whatever diminished health status smokers report because they've had accidents in your model is going to get attributed to smoking?
MR. HAMLIN: Objection, asked and answered.
THE COURT: You may answer that.
A. Well I have not looked at any data set from my examination of the data where that appears to be a reasonable characterization.
Q. When what appears to be a reasonable characterization, doctor?
A. The one you just made.
Q. The characterization that smokers have more accidents?
A. No, that everything is identical except for that one fact.
Q. All right. Well then let's assume that the smokers have more accidents, have more poisonings, exercise less, that they drink more and they have poorer dietary habits and they're more likely to be depressed. Now we've got more differences. All of those differences in your model, if they exist in the data, are going to get attributed to smoking; aren't they?
A. No. The way the model works is as I've stated. To the extent that smokers and similar non-smokers have different expenditures, after structuring for disease in the way we did and looking at similarities in the way I've described, to the extent that smokers had more expenditures than similar non- smokers, those will contribute to smoking-attributable expenditures, and to the extent the reverse is true, it will lower the estimates.
Q. So if smokers who are the same age and the same sex and the same seatbelt usage and other factors that you do include in your model are compared to non-smokers, and the smokers report their health status as lower because they've had more car accidents, that will get attributed to smoking; right?
A. I'm sorry, I didn't follow that.
Q. If smokers who are similar, as you put it, to non- smokers, taking into account what your model takes into account, factors such as age and their sex and whether they wear their seatbelt, okay, have more accidents than non- smokers and report their health status as lower because they've had more accidents, that difference will get attributed to their smoking; right?
*26 MR. HAMLIN: Objection, asked and answered.
THE COURT: Do you understand the question?
THE WITNESS: I'm not sure.
THE COURT: Can you rephrase it, counsel.
Q. Let me try it this way. Accidents are not a factor in your self-reported health status model; right? It's not in the variables that you include.
A. It's not one of the specific factors on the list that's used to structure the model.
Q. And poisonings are not a factor in that model; right?
MR. HAMLIN: Objection, Your Honor, asked and answered.
THE COURT: Yeah, we're starting to go over this several times now, counsel.
MR. BIERSTEKER: Well I'm trying to make it clear --
THE COURT: All right. All right.
MR. BIERSTEKER: -- since this witness was having trouble with the question, Your Honor. I don't know what more I can do.
THE COURT: All right. Go ahead.
Q. Doctor, poisonings are not a factor that you've included?
A. Only to the extent that I've said several times.
Q. Well is it --
Is it a variable or isn't it?
A. It's not an explicit factor in one of the formulas, but it is included in the way that I've already stated several times.
Q. Well that's what I'm trying to understand, is the way that it's included. So --
MR. HAMLIN: Objection to counsel's statement, Your Honor.
MR. BIERSTEKER: I -- I apologize.
Q. Is disability a factor that you include in your self- reported health status model?
A. The same answer.
Q. So it's not a variable in the model explicitly.
A. It is not explicitly, and there would be no reason to do that. Disability is a consequence and related to smoking-related conditions, or could be.
Q. And it's related to car accidents too; right?
A. It's a possibility.
Q. All right. And occupation is not one of the variables or factors that you include in the self-reported health status; right?
A. Well it's included, again, in the sense that I've expressed on all these other factors.
Q. It's not an explicit factor; is that correct?
MR. HAMLIN: Objection, asked and answered.
THE COURT: I think it's been answered.
Q. And if smokers differ with respect to non-smokers with regard to any of those factors that are not explicitly considered in your model, the self- reported health status that may be lower as a result of those conditions is going to get blamed on to smoking; isn't it?
A. The model doesn't assess blame.
Q. It will be -- I'm sorry.
MR. HAMLIN: Objection. He just interrupted the witness again.
THE COURT: Were you finished?
THE WITNESS: Yes.
Q. It will get attributed to smoking, won't it?
A. I'm sorry, now what's the question?
Q. To the extent that smokers report diminished health status because they are disabled or because they have an occupation that -- that makes them injured or because they have car accidents or because they get poisoned or because they have a poor mental state, if all those factors are not explicitly included in your model, your model will attribute to their smoking; right?
*27 MR. HAMLIN: Objection, Your Honor, asked and answered. This really has gone on now for quite some time.
MR. BIERSTEKER: Last question along this line, Your Honor?
THE COURT: All right. Go ahead.
MR. BIERSTEKER: Thank you.
A. Again, to the extent that any conditions by smokers or non-smokers within one of our groups which are set up in a reasonable way to compare like to like on many factors, the only way that those conditions or any of those conditions can contribute to smoking-attributable expenditures if -- is if the smokers in the groups have on average higher costs for those conditions than non- smokers in the same groups. And in reverse as well.
Q. All right. And -- and you don't know whether smokers have higher costs for those conditions in the same groups or not because you didn't look; right?
A. What we did is what I said, we made the calculation, and comparing smokers and non-smokers on this basis is what resulted in a smoking- attributable expenditure estimate.
Q. Now approximately 720 million of the 952 for diminished health status is for 19- to 34-year-old males; right?
A. No, that's a mischaracterization.
Q. Isn't that the estimate that you get from the 19- to 34- year-old males?
A. I believe it's a mischaracterization, going back to your previous question, because we are not making estimates for 19- to 34-year-old males. That is not an estimate.
Q. Do you make an estimate for 19- to 34-year-olds?
A. No.
Q. Do you make an estimate for males over the age -- for anybody, for people over the age of 65?
A. What we're --
What we do in the model is look at estimates for populations.
Q. I understand, doctor, but do you have a estimate for males over the age of 65 -- or excuse me, people over the age of 65?
MR. HAMLIN: Objection, Your Honor, asked and answered.
THE COURT: You may answer.
A. We have made calculations of what the model says for a group of people over the age of 65, yes.
Q. And you have an estimate, then, for over the age of 65; right?
A. We have calculated what the model says about those people, yes.
Q. Okay. And you have dollars associated with those people; right?
A. No --
Well, there are obviously dollars in the model associated with those people.
Q. Okay. And you do the same thing for the 19- to 34-year- olds; right?
A. Do what same thing?
Q. You estimate the smoking-attributable expenditures and the percentage of their expenditures that are attributable to their smoking; right?
A. I've not made any estimate for 19- to 34-year-old males.
Q. I asked about --
I think the question was, and I apologize if I misspoke, but did you make an estimate for 19- to 34-year-olds, whether they're males or females or together?
A. I don't believe I've done that.
Q. But you do have an estimate for people over 65?
MR. HAMLIN: Objection, asked and answered.
A. I have made a calculation of that sort for people over 65 as a means of looking at the model.
*28 Q. And you did that in order to make a comparison to the Medicare study that you displayed to the jury yesterday; right?
A. That's correct.
Q. And you can make the same kind of comparison --
You could make the same kind of calculation for the 19- to 34-year-olds; right?
A. One could do that. But if it's a smaller subgroup, it's not clear without looking at the statistics whether that's a reasonable thing to do or not.
Q. And so you don't know whether 720 million dollars of your 952 million for diminished health status is among males 19 to 34; is that right?
A. Males 19 to 34 make some contribution.
Q. Do you know what that contribution is?
A. That contribution is in the hundreds of millions of dollars, and it's offset by hundreds of millions of dollars in offsets from other groups in the same categories.
Q. Do you know what percentage of males 19 to 34 are eligible for Medicaid because they're disabled?
A. No.
Q. Don't you think that would be important to know if you're going to try to estimate the self-reported health status of males 19 to 34?
A. I think that is not important to know in the context of reasonable statistical methods to be applied in situations such as this one.
Q. So you don't think it was reasonable to include that factor even though you applied the results from your national data set to Minnesota Medicaid; right?
MR. HAMLIN: Objection, asked and answered.
THE COURT: You may answer that.
A. I'm sorry, what was the question?
Q. You didn't think that was important to know even though you take the results from your national data set and apply them to 19- to 34-year-old males on Medicaid here in Minnesota; right?
A. When we apply the national data, we take into account things such as Medicaid status.
Q. You don't take into account Medicaid status when you estimate their health status; do you?
A. No.
Q. Do you think the relationship between smoking and someone's self- reported health status is going to be different in a population of people who live and work in the community than it would be in a group of people who are disabled?
A. It may be. But there are consistent smoking effects adverse on health, is my understanding from the foundation laid by Dr. Samet, and to the extent those are different, there are factors in our model that would adjust for that.
Q. What factors are those, doctor?
A. Education.
Q. Education adjusts for disability; is that --
A. No.
Q. -- your testimony?
A. No, that was not my testimony.
Q. Did you compare people on public aid and the people who weren't on public aid to determine how they might be different?
A. I have made some comparisons like that.
Q. Do people on public aid exercise less?
A. That I don't know.
Q. Do they have more backaches and headaches?
A. That I don't know either.
Q. Are women on public aid less likely to get a mammogram?
A. I don't know.
Q. Are women on public aid more likely to feel down or depressed?
*29 A. I have not made that calculation.
Q. Are people on public aid more likely to have trouble walking one block or climbing stairs?
A. I don't know.
Q. Are women on public aid more likely not to have had a breast examination in the last two years?
A. What I know is that women on public aid tend to be smokers and tend to have poorer health.
Q. Do you know if people on public aid are more likely to have car accidents?
A. No.
Q. Do you know if people on public aid are more likely to have injuries other than those caused by car accidents?
A. We've taken into account those factors, again, in all the models in exactly the way I've stated.
Q. And that was with education?
A. Comparing likes to like.
Q. But you haven't compared like to like with respect to these factors; have you?
MR. HAMLIN: Objection, asked and answered.
THE COURT: You may answer that.
A. I believe we compared likes to like according to standard factors that to a reasonable degree would make the smokers and non- smokers similar in our comparisons.
Q. Well doctor, which of the factors that you did include in your model takes into account different rates of injuries?
A. One of the purposes of controlling for many of the factors that we did control for is to take into account the fact that many of these conditions are related to the factors explicitly in our model, and the only way beyond that that they can contribute to smoking-attributable expenditures is -- is if smokers, who are similar in all these other characteristics, have more expenditures on average than non-smokers.
MR. BIERSTEKER: Could I have the question read back, please.
(Record read by the court reporter.)
MR. BIERSTEKER: Your Honor, I'd move to strike and ask that I get an answer to that question.
MR. HAMLIN: Your Honor, that was responsive to the question which had been asked before.
THE COURT: Okay, I'll let the answer stand.
BY MR. BIERSTEKER:
Q. Do you know whether people on public aid are more likely to be drug dependent, more likely to be alcohol dependent, more likely to have mental impairments, than people who are not on public aid?
A. I've not made that calculation.
Q. Is the estimate that you put up on the board for the 65- plus people yesterday, is that your best estimate for people who are over the age of 65?
A. Yes.
Q. And whatever estimate you got for people who were 19 to 34, is that your best estimate for those people too?
A. We're talking about 19- to 34-year-old males?
Q. Certainly we can talk about 19- to 34-year-old males, that's fine.
A. No. In looking at the data of that subgroup, I would make additional adjustments if I were going to estimate for 19- to 34- year-old males because of the pattern of offsets and fluctuations that you see in the data. You want to take into account the offsets in an appropriate manner.
Q. Okay. So the estimate that you get for 19- to 34-year-old males is not the best estimate; is that right?
*30 MR. HAMLIN: Objection, asked and answered. That misconstrues the witness's testimony.
THE COURT: You may answer that.
A. What we make use of for 19- to 34-year-old males is their contribution in an appropriate way to the calculation of total smoking-attributable expenditures, and one that allows in an appropriate way for offsets in reliance on the law of averages to result in a reasonable estimate.
Q. So it is a reasonable estimate, then, for males 19 to 34.
MR. HAMLIN: Objection, asked and answered, and that mischaracterizes the witness's testimony again.
THE COURT: I'll allow you to answer that.
A. What is in the calculation of total smoking-attributable expenditures from 19- to 34-year-old males is a reasonable calculation of their contribution to an overall estimate for the population.
Q. And is it a best estimate for that group?
MR. HAMLIN: Objection, asked and answered. Your Honor, we have been over this several times.
THE COURT: I think it's been answered now. Counsel, we'll take a short recess.
MR. BIERSTEKER: That will be fine. For lunch?
THE COURT: We aren't going to go through this again, are we?
(Laughter.)
THE COURT: Why don't we recess, reconvene at 1:30.
THE CLERK: Court stands in recess to reconvene at 1:30.
(Recess taken.)
*1 TITLE: STATE OF MINNESOTA AND BLUE CROSS AND BLUE SHIELD OF MINNESOTA, PLAINTIFFS, V. PHILIP MORRIS, INC., ET. AL., DEFENDANTS.
TOPIC: TRIAL TRANSCRIPT
TRANSCRIPT OF PROCEEDINGS
DOCKET-NUMBER: C1-94-8565
VENUE: Minnesota District Court, Second Judicial District, Ramsey County.
YEAR: February 27, 1998
P.M. Session
JUDGE: Hon. Judge Kenneth J. Fitzpatrick, Chief Judge
TEXT:
AFTERNOON SESSION.
THE CLERK: All rise. Court is again in session.
(Jury enters the courtroom.)
THE CLERK: Please be seated.
THE COURT: Counsel.
MR. BIERSTEKER: Thank you, Your Honor.
Good afternoon.
(Collective "Good afternoon.")
BY MR. BIERSTEKER:
Q. Doctor, I'd like to finish off self-reported health status, if we could.
Do you know how many 19- to 34-year-old male smokers on public aid were in the hospital in your national sample?
A. I don't remember exactly.
Q. Do you know whether or not your model attributes 90 percent -- actually more than 90 percent of their hospital care costs to their smoking?
A. Again, in a small group like that I think it's a mischaracterization to say that it attributes to some characteristic like that. It may be --
Q. Do you know --
I'm sorry. I didn't mean to interrupt. Go ahead, I'm sorry. Were you finished?
A. Yes.
Q. I didn't mean to cut you off.
Well do you know what percentage of their expenditures get attributed to smoking and then added into your total?
A. I don't recall that calculation, no.
Q. Do you know why those 26 young men, or however many there were in that group, were in the hospital?
A. I didn't look at those records. There are a variety of reasons, including pneumonia in one case, and many of those people had other smoking- related conditions in their histories that could have led to complications of treatment for other conditions, as Dr. Samet has testified. That's pretty much the extent of my investigation into that.
Q. Do you know whether one of those young men was a kidney donor?
A. I believe that's correct. Those are the kinds of things that are taken into account by the third reduction, the extent that there are kidney donors on one side or the other, and all the other kinds of conditions like that, many of which occur in small groups only among smokers and many of which only among non-smokers. The way the model works is to go over all groups and partition them out in a fair way.
Q. Were there any kidney donors among the non-smokers?
A. I don't recall a kidney donor among the non-smokers, just as in a neighboring group there was cerebral palsy among the non- smokers, none among the smokers.
Q. Do you know whether one of those young men was in the hospital for hemorrhoids?
*2 A. It's possible.
Q. Do you know whether one of those young men was in the hospital for schizophrenia?
A. It could be. But again, I don't believe that that makes any difference to the fundamental workings of the model. It takes all such conditions and parcels them out in an equitable way based on the data.
Q. And assigns them to smoking unless the distribution of hemorrhoids and kidney donors and schizophrenia is the same among non-smokers and smokers; right?
A. Or -- excuse me.
MR. HAMLIN: Objection.
A. Or assigns to --
MR. HAMLIN: Objection, assuming facts not in evidence, foundation.
THE COURT: Counsel, don't interrupt your own witness.
MR. HAMLIN: I thought that he was waiting for me because he saw that I was getting up. But I will not do that, Your Honor.
THE COURT: Do you want to finish your answer, or do you want to rely on your counsel's interruption?
(Laughter.)
THE WITNESS: At this point I'm not sure where I stand in the answer anyway.
(Laughter.)
THE COURT: Well I can't give you legal advice.
THE WITNESS: Where are we? I'm sorry.
THE COURT: Why don't you just ask another question.
MR. BIERSTEKER: All right.
BY MR. BIERSTEKER:
Q. Your model takes into account imbalances -- I mean it takes into account things like schizophrenia or kidney donors or hemorrhoids, and if there's an equal number of schizophrenia and hemorrhoids and kidney donors among the non- smokers and the smokers, it comes out in the wash; right?
A. That --
MR. HAMLIN: Objection, vague.
THE COURT: Well you may answer that.
A. I'm not quite sure how to answer that.
Q. Well if there are more kidney donors and more schizophrenics and more people with hemorrhoids and more people with drug addiction and more people with mental disorders among the smokers, those dollars get attributed to their smoking; right?
A. The way it works, first of all, is not on counts of how things happen, it works on total expenditures, and one reason that's important is because of the complications of treatment of these conditions that can occur when you have smoking-related illnesses, and what the model looks at is the balance of total expenditures for smokers versus similar non-smokers.
Q. Would it be reasonable, doctor, to attribute over 90 percent of the costs of somebody going in the hospital to donate their kidney to their smoking because of a complication?
A. The model does not address the costs of any individual person in the model. Individual persons make contributions, and what happens in the category we're talking about, that the net sum of those contributions, and for totally distinct groups even, comes to four to five percent of the total dollars spent after balancing out all of these kinds of conditions you're talking about from smokers to non-smokers.
Q. Well let me ask you this question, then, doctor. You showed to the jury earlier today, and we'll follow it up later, relative errors for your different calculations. Do you remember that?
*3 A. Yes.
Q. Did you compute --
And one of the relative errors you had was a relative error for the diminished health status mixed catagory; correct?
A. That's correct.
Q. And I believe it was 154 percent?
A. I believe that's right.
Q. Have you calculated --
And the relative error is a measure of uncertainty; right?
A. Of what?
Q. Uncertainty.
A. Well it's a measure of variability.
Q. Variability.
Did you calculate a relative error for your 19- to 34-year- old males?
A. It may be part of the calculations in terms of the contribution, but what we've seen exactly is that there is variation in the groups where the 19- to 34-year-old males are, which is precisely the groups where you would expect to see fluctuation from one group to anoth