Department of Health and Human Services

October 19-20, 2000

Day One

Conference Center

Hilton Palacio del Rio

San Antonio, Texas

ATTENDANCE

Robert W. Harrison, M.D., University of Rochester, Chairman

Michael A. Stoto, Ph.D., George Washington University

Michael Gough, M.D.

Robert C. Stills, Ph.D., NIEHS

Paul R. Camacho, Ph.D., University of Massachusetts-Boston

Steve Selvin, Ph.D., University of California-Berkeley

Ronald F. Coene, P.E., Deputy Director, NCTR,

Exec Sec of the Committee

Barbara Jewell, NCTR, staff

COL Harry E. Marden, M.D., Brooks Air Force Base

LTC Karen A. Fox, M.D., Brooks Air Force Base

LTC Bruce Burnham, Chief of Population Research

Dr. Joel Michalek, Principal Investigator

Debbie del Junco, UTSPH/Houston

Angela Garzon, UTSPH/Houston

Dr. Judson Miner, Program Management Support

Manuel A. Blanca, Program Management Support

Meghan Yeager, SAIC

William Grubbs, SAIC

MAJ Jack Spey (Ret.) Ranch Hands Vietnam Association

C O N T E N T S

Ranch Hand Study Overview

Dr. Michalek 11

Contracting/Program Management overview

Dr. Miner 141

Review of Minutes 154

Action Items 158

Institute of Medicine and Environmental Protection Agency reports 205

NIH/NIEHS interface - RFP process 258

Discussion of Mechanisms for Additional/New Research 249

Review Proposals for Research 439

P R O C E E D I N G S

[8:11 a.m.]

DR. HARRISON: Good morning. I'm Bob Harrison, chairing this session. We are missing one committee member, Mike Stoto, who like Elvis, we know is somewhere in the building -- he's been sighted.

(Laughter)

-- and we assume that he's somewhere down in the basement or subbasement, wandering around wondering where the conference rooms are. And he'll eventually ask someone and show up.

In the meantime, though, I thought we'd get started, and first of all I'd like to say how happy I am to still be a participant with the Air Force in this, what I think is one of the more interesting health studies that I'm aware of -- but then I'm relatively unaware of most things.

At the beginning, because we have a few new players I think on both sides, maybe we could start off by just a two sentence statement about ourselves; and just go around the entire table, include everyone unless you deliberately wish to remain obscure.

I'm Robert Harrison, I'm Professor of Medicine at the University of Rochester in Rochester, New York. My scientific interest is in the mechanisms of action of steroid hormones, but I do a significant amount of clinical care and clinical research.

We'll just go around this way.

DR. GOUGH: Okay, I'm Mike --

[Dr. Stoto arrives.]

DR. STOTO: I got carried away on the Riverwalk this morning.

DR. GOUGH: I'm Michael Gough. First of all, I want to say I was here for five years and when I was here before, we brought Bob Harrison on to straighten us out about the relationship between dioxin and diabetes, and I read in the minutes from the last meeting you hadn't done that yet.

(Laughter)

What's going on?

DR. HARRISON: It just shows, you can fool some of the people some of the time --

(Laughter)

DR. GOUGH: Anyway, I'm a semiretired consultant and I've been involved with Agent Orange since 1980.

DR. SILLS: My name is Robert Sills. I'm a pathologist with the National Institute of Environmental Health Sciences. My research is in carcinogenesis and toxicology studies with the National Toxicology program.

DR. CAMACHO: My name is Paul Camacho, and I'm with the William Joiner Center at the University of Massachusetts. I'm a sociologist, I'm into surveys and a lot of MIS and IS. I'm trying to do decision-making systems, computer systems.

LTC BURNHAM: I'm Bruce Burnham, a veterinarian with a Master's in Public Health, and I'm the military face on the scientific side. Generally we rotate through every three years, and I've been here for a year now.

LTC FOX: I'm Dr. Karen Fox, I'm with the Air Force, and I'm an occupational medicine physician. And I'm getting involved. I don't know if I'm representing Colonel Marden or not, because I expected him to be here, but I may be doing that. I work for him.

DR. MICHALEK: I'm Joel Michalek, Principal Investigator of the study. I have a doctorate in mathematical statistics. I've been with the study since the beginning, 1976.

DR. MINER: I'm Jay Miner, former principal investigator, been with the study since 1985; after I retired from active duty I came back to the program management side of the house, and I'm a contractor now doing acquisition support, making sure that all the science that Dr. Michalek wants gets on contract and gets done.

DR. SELVIN: I'm Steve Selvin, from the University of California at Berkeley. I'm a biostatistician-epidemiologist, and I've been on the project about 15 minutes.

(Laughter)

DR. STOTO: I've been here less than that this morning; but I'm Mike Stoto from George Washington University. I'm an epidemiologist and biostatistician as well. Before I had that job I worked at the University of Medicine and did a lot of the Agent Orange work there, too. So I'm involved with the study from that perspective.

MS. JEWELL: Barbara Jewell with FDA, and I work with the advisory committee, with Ron.

MR. COENE: I'm Ron Coene, and I'm the Deputy Director for the National Center for Toxicological Research of the Food and Drug Administration. And I serve as the Executive Secretary to this committee. For a couple of you who are new, back in '79 the Department was named to oversight this committee, oversight this study, and they passed the baton around the various components of the Department of Health and Human Services to support this function.

So that people wonder why it's I the Food and Drug Administration; well, ten years ago, eleven years ago -- eleven years ago I had a director who was on the sixth floor of the HHS building who said "Sure, we'll do it." And that's how it ended up at the National Center for Toxicological Research. So we've been at it since '89.

DR. HARRISON: How about going back down the wall this way, then?

MAJ SPEY: My name is Jack Spey, I'm a retired Major, President of the Ranch Hand Vietnam Association. I served over there for three and a half years and I've worked real closely with the members of the Air Force Health Study.

MS. YEAGER: I'm Meghan Yeager, from SAIC.

DR. GRUBBS: Bill Grubbs, SAIC. I've been supporting Dr. Michalek and the Ranch Hand Program since 1985.

DR. JACKSON: I'm Billy Jackson, a statistician who works for Dr. Michalek.

MR. BLANCAS: And I'm Manny Blancas, I'm a contractor working alongside Dr. Miner on the program management side of the house.

MR. COENE: All right. We're here, we'll be passing around, if you haven't already done so, a sign-in sheet so we duly record all of you here, as it is a public, open meeting. We don't, other than Jack -- you're the only public we have, again.

You should know you have a whole hour on the agenda tomorrow.

(Laughter)

DR. HARRISON: Well, at any rate, we're glad to have you.

MR. COENE: Thank you for showing interest and being here.

DR. HARRISON: Okay. So we're complete, we're ready?

Joel, it's in your hands.

[Slide]

DR. MICHALEK: Good morning, members of the committee, and friends. I'm Joel Michalek, and this is an overview of the Ranch Hands study. I estimate this will take approximately 45 minutes to an hour; and I'm expecting that people will interrupt me and ask questions, because that's the best way to present this material.

Can everyone hear me okay?

MR. COENE: You're fine, Joel.

DR. MICHALEK: I need to stop right here and talk about the opening slide a little bit. You probably ought to know this; the official name of the study and the protocol is The Air Force Health Study. However, it has another name which everyone knows, the Ranch Hands call it the Ranch Hand study; it's also called in the federal budget, the line item for this study in the federal budget is Ranch Hand II Epidemiology Study. So there are a number of names associated with what we do here.

And the work, by the way, there's a whole raft of people here that are supporting everything you see; they're all back at the base. Some of them are here today. We have really two parts to the organization that make this study work. We have the program managers that keep us funded and keep us legal to all of the just raft of papers associated with contracts and purchase orders and whatever else goes along with committing federal funds, and then there's the group that actually conducts the study; that's where I work. This is represented on the last two lines.

And by the way, I have handouts.

[Passing documents out] Every slide that I'm going to show is on those sheets.

And what we'd like to do when this is over is to put all these slides on our web page so you can get to them back home anytime you want to.

Here comes Colonel Marden, another investigator in the study.

What I have to do is two things: First, you have to know that what I'm trying to do here is get you to put your arms around the whole thing; and then later today and tomorrow we'll get into some topics in much greater detail. So we're going to be touching on things here very lightly just so you will see everything. And in so doing, I have to cover some things very lightly; but I'm sure you'll have questions, so you're free, of course, to stop me anytime, to make a note and get me later or send me an e-mail after this is all over with, or whatever you want to do, and we can operate that way.

[Slide]

So why are we here? We sprayed approximately 19 million gallons of herbicide in Vietnam between 1961 and 1971. That led to concern by veterans subsequent to the war; I think the key date was sometime in 1975, a claims clerk at the VA Hospital in Chicago called the newspapers to report her concern that she was seeing excess symptoms in Vietnam veterans.

That led to a lot of things, as you recall from those years; it led to in particular congressional hearings. A key hearing in 1980, attended by the Deputy Surgeon General of the Air Force, statements were made by him to be responsive and committal of Air Force resources to study the issue in the men that sprayed this material in Vietnam, the Ranch Hand veterans.

And it was from that point forward that we mark the beginning of the study. Actually, prior to that phone calls were made to Brooks Air Force Base for us to begin contemplating an Agent Orange protocol, and so we began the technical side of the issue of writing a protocol in 1976 -- but I'm getting ahead of things right here.

Here's a slide, Contents, that we're going to talk about; why are we here and where did all this come from? With an overview of Results, an overview of our publications, and the recent GAO report, recently expressed veterans frustrations that we're going to talk about, and how we can address those. And some suggestions on those frustrations right there, and some pictures of our facilities.

[Slide]

So here's the goal, of course. The epidemiologic template was applied; namely to ask the question, did we harm any of our Ranch Hand veterans in any way -- and 'any way' means health, mortality and reproductive outcomes, by means of their spraying of the herbicide which we found out, subsequent, after the war -- actually late in the war -- much of it was contaminated with dioxin, also called TCDD.

[Slide]

Here's a slide representative the key documents that launched this study. There was a letter from the White House to the Secretary of Defense, I believe, or Secretary of the Air Force directing the Department of Defense to conduct this study. And that letter is in my file, dated 16 September 1980, from Stuart Eisenstadt, Domestic Policy Counsel to the President.

Prior to that of course were the hearings that I have already described. That produced a funding element in the federal budget specifically devoted to this study; and that, by the way, is the reason why we're all here today, that this study has dedicated funding. And we all know how hard it is to maintain funding over a long period of time in any study, but we've been fortunate in that regard.

Since then there's been a public law in 1990 to structure the committee that we see here today to allow and ensure veteran participation.

[Slide]

We have points of contact at the Pentagon in Washington and at Brooks Air Force Base regarding our contracting. And I have a pointer.

[Slide]

We have, I already talked about our program mangers; and you'll see some pictures later on. We have about 30 people, 35 people working on the study today of which ten are civil service, two active duty military, one of which is right here. Programmers, statisticians, medical coders, scanners, student aides, whatever it takes to do a study of this scope and duration.

[Slide]

It's a multifaceted operation. Here's us, here's my group right here in Brooks Air Force Base, technical side, and here's the managers you see here today. It's an enormous effort. It could not be done at Brooks Air Force Base by us; the physical examinations I'm going to talk about are done in California, those are overseen by our prime contractor, SAIC Corporation, which is right here, and those physical exams are conducted at Scripps Clinic and the interviewing is done by the National Opinion Research Center of the University of Chicago. Those are all subcontractors to Science Applications International Corporation.

Those contracts are managed by these two individuals who are sitting here today, Dr. Jay Miner and Manny Blancas, along with Major Kyle Sneddon.

Then our technical side of the study, I have myself, the other statisticians of which one is here today, the other investigators, back up here we have Program Support, it

interfaces with Congress and our money and funding at the Pentagon, and we have interface with CDC, NIH, the EPA and NIEHS, among others, and Department of Veterans Affairs, other government agencies we talk to and communicate with all the time about our study.

And we have our Advisory Committee that's sitting with us today, and other support contractors that keep us going in-house; namely our statisticians and scanners.

To date we have spent about $150 million on this study, since the beginning.

[Slide]

Well, here is the beginning. Roughly 1976, 1977 -- that was when we were asked to drop everything and begin to concentrate on writing an Agent Orange protocol. There was a peer review process that took place between 1977 and 1981, before the National Academy of Sciences, the Armed Forces Epidemiology Board, the Air Force's Scientific Advisory Board, and others during that period to refine the protocol which was made final around it, beginning of 1980-'81, and that was the basis for the first physical examination, that occurred in 1982 at Kelsey Seybold Clinic in Houston.

The protocol is available on our web page. In fact, almost everything I'm talking about today is on our web page, and the web page address is on the last slide of your handout.

Protocol, as you see, called for periodic physical examinations of the study subjects and their controls roughly every five years with a sort of break in the pattern here in '85. The pattern was '82, '87, '92, '97, and the last physical is programmed for the year 2002. Now we had an extra physical here in '85.

I think it's important and significant but often forgotten that the environment in which this study was contemplated was one of fear. The fear was that not only had we lost the war in Vietnam, that we had poisoned our own troops. That fear is represented right here.

I can't communicate very well except the following way: In 1984, we gave our first press conference at the Pentagon on our first mortality study of the Ranch Hand unit. At that time the overall Ranch Hand -- and since then, in fact, the overall mortality is nearly identical, Ranch Handers to the controls.

And we have a video of this press conference. It was a room three times this big, just packed with television crews, newspapers, lights, public, commotion, talk, and I'm up here presenting slides like I am today of the results of our first mortality study.

So to a statistician it's pure vanilla, hardly anything happening. Relative risk 1.0, back in '84. We were showing a Kaplan-Maier survival curve, which was probably too sophisticated and too detailed for that audience, but I didn't know that at the time. So I'm showing a Kaplan-Maier survival curve. The Kaplan-Maier curve has little steps, smooth, comes down. Every step is created by a death; that's how steps occur.

So someone from the audience, after I gave the results, asked: What's that little jump in the curve right there? Why does it go in those little steps? I said "Well, that's because somebody died." And as soon as I said the word die, the room fell silent. You could hear a pin drop. There was no overall effect. The relative risk was 1.0, but that's the environment we were in in 1984. That's the environment we were in in 1976 when the hearings took place that led to this study.

That's why we did this extra physical in '85. We expected the Ranch Handers to be expressing acute effects. We wanted to catch it so we could intervene and help them. That's the spirit of this pattern right here; that's where it came from.

In fact, we worry about that. How often should we have these first few physicals, to catch the effect, to intervene if we had to? [Slide]

So what do we have? We have applied the standard epidemiologic template to a problem of unprecedented scope and complication. We've had 1261 Ranch Handers who ever existed. 26 were killed in action, 50 or so had died of natural causes before the first physical examination. There were 1208 eligible for the first physical in 1982.

We had a population of Air Force veterans who were in Southeast Asia during the same time period, but had nothing to do with spraying agent Orange. They were flying C130s and servicing -- they were the air and ground crew for C130 aircraft that were used for all kinds of purposes, such as cargo, air-sea reconnaissance, air-sea rescue, whatever. The C130 was used for a lot of things. It was not used for spraying herbicides.

That's the population of 19,080 that have in our control population. We have a matched design where about 10,000 of those are matched on a one-to-many basis to these Ranch Handers. Matched on date of birth, military occupation, race.

So we have up to 8 to 10 Ranch Handers.

[Two women enter the hearing room.]

This is Debbie del Junco, University of Texas at Houston, and Angela Garzon, one of her students, who we invited.

Matched up to 8 to 10 comparisons per Ranch Hander on those variables that I've just mentioned; officer matched to officer, enlisted flyer to enlisted flyer, and so on.

In each matched set, those individuals were randomized, in random order. And after randomization in the first position, was invited to attend a physical examination in 1982, along with his respective Ranch Hander.

So at the beginning it was designed to be a 1:1 matched design. Subsequent to that, when an individual such as a control refused to come or became noncompliant, he was replaced following a strategy that's defined in the appendix of the protocol.

The idea was, we were afraid that we were going to lose a large proportion of our comparison group due to noncompliance and lack of interest. We expected -- you'll see in our protocol -- we were expected to lose 50 percent of our controls in the first few years of the study.

So with our advisory committees and with approval and through peer review, we devised a replacement strategy such that an individual who becomes noncompliant is replaced by an individual from the same matched set who has the same perception of health as the one who refused.

The refusal says I'm in excellent health; we look for another comparison in the same matched set that reports excellent health. That's the replacement strategy.

If that match can't be made, there's a scale; excellent, good, fair, poor -- then it's dichotomized: excellent to good, fair-poor. Then if we can match them on a dichotomized scale, we'll do so. If we can't match at all, we don't replace.

So it's with that strategy that we built into the protocol, we attempted to compensate for the expected losses in the control group.

Well, it turns out we didn't realize the losses that we expected; the compliance has been very good. But we still have our replacement strategy today.

DR. MINER: Once invited, always back.

DR. MICHALEK: Yes. There are more rules; we don't replace dead controls, and once a control is invited, he's always invited, for the rest of the study. Everyone is always invited back.

[Slide]

So what is the epidemiologic template? In principle it's very simple. You must define an exposed cohort and you must be thorough and you have to ascertain an exposed cohort. You don't want to take the people that walk into a clinic, for example. We have a roster. We know exactly all the Ranch Hands who ever existed. They were all identified and all living Ranch Handers were invited.

And we have a full ascertainment of our comparison group. Separately, the rest of the template is to devise an exposure index within the exposed group. What we're looking for here of course are group differences on health, and within the exposed group we're looking for a trend; we want to see the individuals with high exposure, higher risk than individuals of lower exposure.

The exposure index is the problem, of course. There was no dosimetry in Vietnam. In fact, when the herbicides were sprayed in Vietnam, we thought they were safe. We told them it was safe. No dosimetry.

The issue arose in 1976-77 after the war had ended. In other words, when you have a pattern, we have a scenario here similar to what happened with the Gulf War.

What made this study work, and what made all Agent Orange studies work, is the fact that the contaminant has a very long half life. We can measure it in the blood today, even 30 years after the exposure you can see it in the blood, because the half life is so long, it's so persistent. And it was because of that we were able to construct an exposure index that we had confidence in later on.

But at the time you wrote the protocol, the exposure index was contemplated to be based on military records and gallons sprayed in Vietnam, that I'll talk about in a minute.

The study's unprecedented scope: What was epidemiology before the Ranch Hand study? The classic example was the British article, I guess 1953, Hill and Pito, I guess. -- Hill and Dahl. It was on smoking in physicians in England.

They collected smoking information by questionnaire from physicians in England. And they looked at lung cancer. They had a well-defined exposure, they had a well-defined end point. The results were clear as a bell.

There was a significant trend; that is the classic paper. That is epidemiology. That was epidemiology before this study came along.

What do we have in this study? Number one, we don't know what we're looking for. The veterans were complaining of heart disease, of cancer, anxiety, birth defects, diabetes, skin conditions, you name it. There was a list.

It was that, the list of conditions by the way, came from the Department of Veterans Affairs. That was used to devise and design the first physical examination, to address all of those conditions. That was unprecedented.

Secondly, we didn't even know what the dose was, because we didn't have any data; except that the overall amount of herbicide spray in the whole country, we didn't have any data specific to the individual; we only had global information of the whole country of Vietnam.

So we were in unprecedented territory here. As another example of the environment in which this study was conceived, we nevertheless had the mandate to proceed, and we did. We applied the standard epidemiologic template, and as you'll see, with some great success, because of certain things that happened along the way technologically.

We had multiple endpoints, no believable exposure index at the beginning; and you'll see we applied the standard epidemiologic template for physical examinations, interviews, mortality assessments; and really an unprecedented effort to collect quality information.

[Slide]

Here are some numbers to show you what kind of compliance we've had from the beginning. Here are the number eligible, here are the number that actually complied with the physical examination on the Ranch Hand group and on the comparison group. And you'll see about 80 to 85 percent of the Ranch Hand group have been compliant, and about 75 percent of the control group have been compliant.

This is beyond our expectations when we wrote the protocol, that we would see such compliance. It also puts this study into a round of -- and you have to view a lot of studies -- this is probably one of the best studies ever done, from many points of view.

[Slide]

These men live all over the country. Right now the physical exams are done at the Scripps Clinic, California, first physical done in Houston, Texas. Our prime contractor is in Virginia and in San Diego. And our National Opinion Research Center from Chicago are here at Brooks Air Force Base.

We literally move 2300 men to California every five years. We purchase 7500 room-nights at the La Jolla Hilton every five years. It's a massive effort.

When we conduct physicals, when we move -- transport and physically examine 2300 men at Scripps Clinic, we do that over a ten month period, and we spend about $16 million doing it.

In a year in which we're not doing physicals, we're spending about $5 million on salaries and overhead to support the research activities at Brooks Air Force Base.

[Slide]

These are the words I've already said; no dosimetry, unprecedented scope. We expected great loss to follow up in compliance; that was not realized, fortunately. We have a matched design to replace strategy, and great concern about exposure excess than credibility.

One of the objects -- let's talk about this for a second. This was mentioned at the Shays hearing before the House Government Reform Committee hearing earlier this year. Another point that's important but often forgotten, one of the items on the table in 1977 and '78 was that the Air Force should not do this study, that someone else should do it.

And that was on the table in front of every advisory and overview and peer review committee until the very end, when it was decided by our peer reviewers that the Air Force should conduct the study. Why? Because it was a compelling need that this study be launched immediately and the results be obtained as soon as possible. And the Air Force had the resources and the knowledge to do that.

[Slide]

Design and analysis, there was a lot of argument at the beginning about what the control group should be; and that's still facing us today. Scientifically, of course, what you want as a control group that is the same in every way as the exposed, except for one thing; and that's the exposure, of course.

There was concern about the possibility that if we had a control group that was stationed somewhere else other than Vietnam, then we would hopelessly confound the study with effects of tropical diseases and all the rest that goes along with being stationed in a war zone in the tropics.

So the decision was, we'll use controls that were stationed in Vietnam during the same time period that the Ranch Hand unit was active, and they will be Air Force controls because of the known differences, the subtlety differences, anyway, because the different services. In other words, Army troops were out of the picture here.

So we have a control group of, like I said, Air Force veterans who were in the area during the same time. It was contemplated during that period, and on the table, by the way, to study the control groups stationed in Europe.

Now today, looking back, having attended some meetings on Gulf War, we now realize the benefit of having multiple control groups, and future studies will have multiple control groups. Such as a control group deployed and a control group non-deployed, and a civilian control group. All those things are being talked about today about future studies. When this study was designed in 1976, this was the idea.

[Slide]

We worried about all of these things on the screen here, and I'm going to talk about many of them today.

In the protocol you will see a formulation of an exposure index based on the number of flights that took place during that fellow's tour. The number of gallons sprayed, the number of days on the job, concentration of dioxin in the herbicide. That was the idea written in the protocol.

That idea was immediately discarded, as soon as the study started, because we realized that we didn't have the data. We didn't have data specific to the individual, so this was scrapped. And what we settled on was an index which was simply the number of gallons sprayed in Vietnam during that individual's tour, times the concentration of the contaminant in the herbicide, which we knew, divided by the number of persons on the job, thinking that, as we threw more men on the job, "Well, gosh, the exposure must be increasing."

Well, actually the exposure must be decreasing, because there were more men there and they'd be sharing the same amount of work. But we found out later that's a pretty lousy assumption; that what really happened on the job was when you threw more people on the job, just more people got exposed, that's all. And so by requiring that the exposure index decrease with the number of people on the job, we were probably committing a mistake.

And as you'll see later, we confirmed -- or we have data to support the idea that this is a pretty poor index.

In 1986 we were invited to a meeting at the Office of Science and Technology Policy, sponsored by OSTP, where we met CDC for the first time. Dr. Don Patterson, Larry Neil, and Eric Sampson. They had devised an assay for dioxin in human serum that was as good as and equivalent to assays that had been done before that in adipose tissue for dioxin.

We launched a pilot study; we sent 150 of our Ranch Hand veterans to three Red Cross clinics in the United States and 50 controls. And we measured, and we drew blood, and we measured dioxin in their serum. And that's published in MMWR 1988, I believe.

The study worked. We found a significant increase of dioxin body burden in the Ranch Hand veterans, which number one validates the idea that the Ranch Handers really were exposed, and as you'll see later, validates a lot of things.

So that was our first experience with the new technology, which was to measure the contaminant in the blood of these men. That was a breakthrough.

[Slide]

Here is a picture showing where the Ranch Handers stand relative to other cohorts. Now there's a lot of caveats associated with the picture, which is what I got from CDC. Here are the Ranch Handers right here in this light blue color, here are the controls, and I broke out by the five occupational categories: non-flying officers, flying officers and so on are listed non-fliers.

Here are the subgroups of the Vietnam Experience Study. The Vietnam Experience Study was intended to be a sister study to this one, based on Army troops. The study consisted of a cohort of Army troops that went to Vietnam and had opportunity for exposure, and a cohort of Army troops that didn't go to Vietnam that of course had no exposure.

They used the same physical examination, they were supposed to follow the same drill we did with repeated physicals and questionnaires. That study was stopped after the first physical in 1987. The reason being that when they assayed them for dioxin, they found background levels in both the exposed group and the controls.

Now as we see today, I regret the decision to stop that study because it has contributed to veteran frustration today. But nevertheless there they are.

[Slide]

And here are the Ranch Handers, measured at the same time approximately, 1987, as the veterans in this Vietnam Experience Study.

Here are the individuals in the NIOSH industrial cohort study. Those are men who worked in factories in the United States that made herbicide. Those men were exposed over roughly a 20 or 30 year period working in industrial factories here in the United States, in chemical plants. They actually have higher levels.

Those levels were collected from blood drawn roughly the same time period, 1987, and I have to keep telling you the time period, because remember, as an individual is dosed, he will eliminate the dioxin from his body due to first order kinetics; so the amount in your body today, if you were exposed ten years ago, is less than it was ten years ago.

And here are two other cohorts of German plant workers which are widely published, and New Zealand herbicide sprayers.

Down here are the individuals who were victims of an explosion of a chemical plant in Italy in 1976, at Seveso, where a number of individuals received up to twenty to thirty thousand parts per trillion. And by the way, the highest level in the Ranch Hand group today is about 660 parts-per-trillion. The highest level in the cohort of the NIOSH study -- well, these are medians. The highest level is further out, is about 3,000 parts-per-trillion.

A parts-per-trillion is 10^-12, which is equivalent to 1 second in 32,000 years or 1 dime in a stack of dimes from here to the sun. CDC can measure that level of contaminant in the body with the same level of accuracy that Scripps Clinic measures insulin; in other words, with a cv of about 9 percent. That is a tribute to the chemistry at CDC, as you'll see in a few minutes.

The caveat here is that these measurements on the Seveso victims were made from blood drawn just a few days after the accident. The caveat here is that these measurements were made from blood drawn in 1987, which is up to 15 years after exposure. So you have to remember that when you look at these slides; that these men, especially our Ranch Hand group, had an initial dose we think that ranged up to about 3,000 parts-per-trillion when they were in Vietnam. Which is still only about a tenth or less of the exposure received by the victims of the Seveso accident.

Remember also then, although the levels in Seveso are very high, the cohort is very small. There are three zones in Seveso; Zone A, B, and R. Zone A received the highest levels -- it's not labeled here, but there are a couple hundred individuals in that zone.

An acute effect of exposure to this chemical is chloracne, which is a skin condition that looks a lot like acne but has a different pattern to it. Individuals are here broken out as to whether or not they had chloracne; and there's no well-defined cut point based on dioxin body burden to determine who will get chloracne and who won't.

[Slide]

So what does dioxin distribution look like in the Ranch Hand group? Now here I'm showing the histogram in raw units, which doesn't look very pretty, because it it's so highly skewed, of the distribution on the Ranch Hand side and on the control side. The controls, 99 percent of the controls have less than 10 parts-per-trillion, currently.

This is published in -- and I'll show you some citations on that -- the median or mean is about 5 parts-per-trillion. All of us in this room have about 5 parts-per-trillion in our blood. We get it from breathing smoke from burning trash, from eating certain fish and seafoods; that's the primary source of uptake in the United States, is diet. And you get it primarily in your diet from seafood and from dairy products, and from meat. Anything that has fat in it.

You can also get trace amounts from plastics and paper products. Just by touching a styrofoam cup, you're getting a tiny amount of dioxin in your body. What happens is that all of us are experiencing constant uptake of a tiny amount every day, and at the same time we're experiencing whole body elimination. So we're at kind of a steady state.

Our body burden is going to fluctuate for the rest of our lives; it will gradually increase, and that's published in 1998, showing the data from our control group. In the Ranch Hands group, of course, you see that decreased. Here it is in log units, which is our favorite transformation in statistics, which shows a nice, approximately normal distribution in the Ranch Hand group and control group.

In the Ranch Hand group, the median is 12 parts per trillion.

DR. CAMACHO: Joel, could you go back to the previous slide, please?

Out there in the Ranch Hand population, the people out there at 600 --

DR. MICHALEK: Right.

DR. CAMACHO: What's the end for that?

DR. MICHALEK: Okay, I'll give you some numbers on the Ranch Hand side. The median is 12; 50 percent have less than 12. In other words, half of the Ranch Hand group look like controls as regards their current body burden, which is not a nice fact to have to face statistically, if you're worried about exposure --.

The percentiles, don't have those memorized. There's one individual with 660 parts-per-trillion. I can get you that a little bit later.

DR. CAMACHO: But that's only one or two people.

DR. MICHALEK: One or two people, right.

I'll give you another number: 98 percentile in the control group is 200. Almost all of them are less than 200, in the Ranch Hand group.

I don't have the other percentiles memorized, but I can get those for you.

DR. HARRISON: Joel, what's the molarity of 12 parts-per-trillion?

(Laughter)

DR. MICHALEK: That's a good question. I have a slide on converting whole weight to liquid weight dioxin, but I don't have a slide converting parts per trillion to molarity. But we can get that. Can't get it for you instantly.

DR. HARRISON: You know, I ask that every meeting.

DR. MICHALEK: Sorry, I don't have that conversion memorized.

DR. MINER: You did that, though.

DR. MICHALEK: No, I didn't do that one.

DR. MINER: I gave you that last time.

DR. HARRISON: Just to put this in for one more time, the argument that I have with environmental assessments is that they assume a relationship between the toxin all the way to zero. Whereas in my world, there's a concentration of active material below which you don't see anything.

LTC BURNHAM: You're seeing a threshold.

DR. HARRISON: And I know the EPA's position is that 1 trillionth is just one trillionth as bad as 1. But in my world, one trillionth is the same as zero.

Jay, you say you know what 12 parts per trillion is?

DR. MINER: No, I said -- I copied some conversion factors and brought them to you last time, but I don't have with me now.

DR. MICHALEK: We can get that.

DR. CAMACHO: On the noncompliant and the people who dropped out of the study, is there any standard like survey done to see if they had anything.

DR. MICHALEK: Yes; they're given a noncompliant questionnaire: Why didn't you want to come? No time, no interest, too sick, whatever.

DR. CAMACHO: And they were spread all over the place.

DR. MICHALEK: Yes, they're spread all over, and the are groups equivalent on that. However, that's an important point, in that when the individual says "I can't come, I'm too sick" we pay attention. When we start to see -- one of two things can happen that would make us very worried.

If a great proportion of them couldn't come because they're too sick to come, or if one group was unbalanced with regard to the other in that direction, that statistic is very important to us and we're watching that. It's only a few percent, one or two percent that can't come because they're too ill. But we have ways to find out about them, too, by means of medical record collection.

So we're on to that, yes, and that's all in our reports.

DR. SILLS: Joel, I have one question. Can you go back to the last slide -- the slide before this. You know when you talk about chloracne with the Italy study, I was just wondering, in terms of your Ranch Hand population, did you see any chloracne?

DR. MICHALEK: We have found no chloracne. That's published in the Archives of Environmental Health, 1998.

And that paper took five years to get published. What we did was we went back to medical records that were collected while they were in Vietnam, on every Ranch Hander, and we studied every record. We found only one individual that had any annotation on his record that he was having a skin problem. And we reported that in the article.

So the article then talks about acne. The intent of the article was chloracne, but the study is affirmant because we didn't have any chloracne to study. So no, we didn't see any.

DR. STOTO: It wouldn't possible, though, if you had an 18 year old man with acne that he wouldn't think it was exceptional enough to --

DR. MICHALEK: Absolutely, yes.

Of course, remember, that in Vietnam at the time that the doctor didn't know; he was told the stuff was safe. And probably the whole concept of chloracne wasn't at the top of his mind at that time, in 1963, '64 when this stuff was being sprayed.

Yes, Jack?

MAJ SPEY: I would just make a comment about that. All the flight crew received annual physicals. We were all in the area, in the general age bracket of between 24 and 28 years old. Had any of us started coming down with acne at 28, 24 years old or 18 or 20 years old, I would have been brought to the attention of the flight surgeons; they wouldn't have recognized the difference between chloracne and ordinary acne because it takes a specially-trained dermatologist to be able to make that determination; but it certainly would have been indicated in part of our health records, and it wasn't.

DR. MICHALEK: And we have all those records.

MS. del JUNCO: Joel, in the group of troops that was the Army, the first group that you guys didn't follow anymore, how many dioxin body parts and samples did you analyze?

DR. MICHALEK: Would you say that again, please.

MS. del JUNCO: In the first group, the one before the Ranch Hands, the ground troop veterans, the ones that included the Army and was discontinued, do you have any samples?

LTC BURNHAM: The Vietnam Experience Study.

DR. MICHALEK: Oh, the Vietnam Experience Study. Yes.

MS. del JUNCO: Okay, the Vietnam Experience Study, as you call it. Do you have any actual dioxin --

DR. MICHALEK: Yes.

MS. del JUNCO: How many samples did you analyze in that group?

DR. MICHALEK: I didn't analyze those. Those were done at CDC and those were published.

DR. GOUGH: There were 600 people from Vietnam, and 80 or 100 non-Vietnam comparisons.

MS. del JUNCO: And these were Army and Marines?

DR. GOUGH: No. All Army. The low, medium and high was categorized by the relationship of the reported positions of the Army units to the Agent Orange spraying missions, which is just subject to all kinds of misclassifications.

But the prediction, from the spray missions, is that the Army troops would not have been exposed, because they weren't very close, they were sprayed only rarely, and there's a lot of diffusion of Agent Orange before it got to the ground.

So those results are consistent with the estimates of what the exposure would have been.

DR. HARRISON: And of course, if you're comparing the effects of dioxin, there were other defoliants used by the Army that contain dioxin; so the actual dioxin exposure is probably not easily estimated.

There's something, when I was looking at the minutes last night that I thought about, Joel, and I guess I might as well ask it now.

There was a small population of men who died before the study began. If you were looking for an acute effect of dioxin, those might have been the ones acutely affected. I know you looked at it. I know you looked at it. I'm just asking you, how did you look and what did you find?

DR. MICHALEK: That was published in JAMA in 1990, the very first mortality study. There aren't any group differences, by cause of death. And that's all we can do; we didn't have dioxin levels at the time.

DR. HARRISON: Let me ask it this way: were most of those deaths cardiovascular? Cardiovascular and renal, let's say.

DR. MICHALEK: We saw that effect later. In 1988-'89, we saw an increased risk of cardiovascular death in the enlisted ground crew, which gets our attention, because they have the highest levels.

Didn't see that, we didn't even know to look so carefully in the early years, but I don't remember and we'd have to check that.

DR. HARRISON: Okay.

DR. STOTO: Weren't they mainly automobile accidents and things like that?

DR. MICHALEK: There was some evidence of increased risk of, external-caused events, deaths; yes, in the first few years after Vietnam.

DR. GOUGH: Which has been observed in other veterans of Vietnam --

DR. MICHALEK: Remember, both groups are Vietnam veterans in this case.

DR. GOUGH: And Korea and World War II.

DR. MICHALEK: True, but both -- our control group was in Vietnam, too.

But it wasn't significant, I don't believe. I don't remember. I have to check.

DR. STOTO: My recollection is that a lot of the deaths were of that sort.

DR. MICHALEK: Oh, yes.

DR. STOTO: That's what you would expect for men of that age.

DR. MICHALEK: Yes. At that time many of the deaths were externally caused.

[Slide]

Okay, how good is this dioxin measurement? Well, fortunately we had this pilot study where we sent them to the clinics.

A few months later they were invited to the 1987 physical at Scripps Clinic in California; and at that point they invited everybody to give blood for dioxin. But we still had this cohort that had been to the clinics. 47 of them volunteered again, so we had paired measurements, within a few months apart, on 47 people; which we used to do a standard, a measure of reliability, and the original units or individuals 11 parts, up to 50 parts per trillion, and the coefficient of reliability is 87 percent.

On the log scale -- because that's the unit we use in all of our analyses -- the coefficient of reliability is 96 percent on a scale of 0 to 100, which means 96 percent of the variability in the measurement is due to true differences between people, and only about 4 percent is due to the noise. Which is very good, considering the scale on which CDC is operated on a part per trillion scale.

DR. STOTO: That's reliability with respect to what the persons' dioxin level was at that time.

DR. MICHALEK: Yes.

DR. STOTO: And if you try to extrapolate back, those numbers would be somewhat different.

DR. MICHALEK: I have some things to show you later that I'm so excited about it's hard for me to tell you.

(Laughter)

Fortunately, we have another meeting in December, and I have some data which combines Seveso half-life studies and the Ranch Hand half-life studies that will address what you just said.

Anyway, here we have, in log units, the classic picture that you want to see, this is what you see in textbooks; you expect to see a 45 degree line. When you plot the dioxin level in the pilot study versus the dioxin level measured at Scripps Clinic, or cut from blood from Scripps Clinic, and it's just very tightly in log units, scattered around a 45 degree line, which gives us great confidence in the measurement.

And here it is in original raw units, and you see a pretty good, tight scatter around here, less than 50 parts per trillion, which is the reason for our statement about less than 50 and greater than that is pretty noisy. But we don't analyze original units anyway; we always analyze in log units, so we're happy about that.

[Slide]

What does this measurement have to do with what actually happened in Vietnam on the job? That was the next question, and the very first question at the tops of our mind is at the time. To address that, we sent a quantity to all enlisted Ranch Handers -- there's about 500 of them -- and we questioned them about on the job activities in Vietnam. And we found out what they did in Vietnam by interviewing two Ranch Hand crew chiefs who happened to live in Texas.

Someone had to get in the tank, it was a thousand gallon tank in the back of the plane, it had a dump valve. Someone had to get in the bank, get down on his hands and knees, and grease the valve. And as I was told, the bank is never completely empty.

Some of them used herbicide as a hand cleaner, because they were told it was safe, and because it actually does a very good job of removing grease and oil from your hands.

Some of them got sprayed in the face and torso as they were standing on landers behind the trailing edge of the wing, sticking coat hangers and screwdrivers into the nozzles, to clear the nozzles. Because the herbicide would dissolve rubber, and so there were little bits of rubber and other crap in the line, it would clog up the nozzle.

They were in tropical heat. This was on the job exposure. And of course they would get herbicide on their clothing. This is a different scenario from the flyers, who didn't receive this kind of exposure because you didn't work in the tanks and fill the tanks like the enlisted; is that true?

MAJ SPEY: Can a make a point?

DR. MICHALEK: Yes.

MAJ SPEY: Just a simple observation. when a flight crew member, a pilot, preflighted his airplane, he walked around the exterior of course; and then when you walk through the cargo compartment, you'd grab a pressure line, you'd touch the tank, you'd check to make sure that the tank cap was on tight. You had scudge on your hands, and then you might wipe the sweat off your face or scratch your eye --

(Laughter)

-- helmet on, or some of us, you know did this. The material was everywhere. I mean, it wasn't wasted, it wasn't flowing across the cargo compartment of the airplane, but anyone that went on that airplane; passenger, crew member or whatever, came in physical contact with the material.

DR. HARRISON: You know, that's something I hadn't thought about, but this stuff is somewhat volatile, isn't it?

DR. MICHALEK: I don't know what that statistic is. The vapor pressure of the herbicide?

DR. HARRISON: In other words --

LTC BURNHAM: You can smell it.

DR. HARRISON: How much did you breathe?

DR. MICHALEK: I don't know those numbers.

DR. HARRISON: And that would be good absorption.

MAJ SPEY: I'm not sure if protocol allows me to answer questions.

DR. HARRISON: It doesn't, but why don't you go ahead, sir?

MAJ SPEY: The air flow in the C-123 -- we flew with the open troop jump doors -- the troop jump doors open so that the flight engineer could pull a pin on a smoke grenade and throw it out to mark the position of ground fire. The front windows in the cockpit were open to prevent shattered plexiglas from injuring us, should a bullet hit that window. Plus, it was our air conditioning system.

(Laughter)

The air flow came in, the troop jump doors in the rear of the aircraft, the odor, et cetera, et cetera, came forward across the inboard side of the face of the pilot and copilot and out that window. You were smelling it all the time. And you know, it smells terrible.

DR. MICHALEK: In retrospect, we should have given that questionnaire to the flyers, but we didn't, to the officers. We only gave the questionnaire to the enlisted. That was because the data at the time showed the enlisted had much higher levels. So that's why we did what we did.

DR. STILLS: Joel, I have one question: In terms of, you mentioned that 66 parts per trillion was the highest exposure that you --

DR. MICHALEK: We saw it in the Ranch Hand group.

DR. SILLS: -- that you saw in this study.

Did you have a nice correlation when you looked at, for example, it's the 66 parts per trillion, was that observed in the men entering the spray tank? Is that where you saw most of --

DR. MICHALEK: We're getting to the next slide, yes.

[Slide]

So here are the activities that were reported to us and which were included in the questionnaire. Here are the results.

We actually looked at the questionnaire and we scored the total number of days of skin exposure, and across the vertical we have dioxin levels in log units -- this is the right hand side; all of the individuals are here. And then we broke the cohort down into categories.

We didn't administer a questionnaire to controls. We included the controls here as a reference. So these are enlisted controls. We have the same experience as the Ranch Hand group, they're the same rank, same activities, but they weren't spraying herbicide.

Then we took this cohort that received the questionnaire, and we broke them out into five categories. Some of them reported being administrators, which meant they sat in an office in the command section, and weren't out on the flight line.

Some of them reported no exposure whatsoever: "I never touched it" and they'd leave their questionnaire blank. And then after that we had the group that reported exposure by means of all the methods you saw in the previous slide.

We broke those out into tertiles, by the number of days of skin exposure. And we looked at that versus their dioxin body burden measured in 1987. I want to use the word "awesome" but this is a technical discussion, so I won't.

This is it. This is the connection between what we measure today and what actually happened in Vietnam. We see this. This validates the dioxin body burden as a measure. It's not perfect, because you see we have individuals here -- that one is almost zero parts per trillion, who had, according to the questionnaire, very high skin exposure.

DR. GOUGH: Are they very skinny?

DR. MICHALEK: Yes. There was a range of percent body fats, percent body fat in Vietnam. And that turns out to be a very important predictor of a lot of things, which I'll talk about in a few minutes. You couldn't be too heavy, because you had to get in the tank, and it was an 18 inch hatch.

Here we look at the flight engineers who operated equipment in flight, and this is the ground crew that filled the tanks, and this is everybody.

This was published in the Journal of Exposure Analysis, 1996 I believe. Somewhere in the Nineties.

[Slide]

Here it is again; here I've simply created a few categories; the administrators, the enlisted flight engineers, enlisted ground crew, showing the high correlation between activities in Vietnam and subsequent body burden of dioxin in log units.

[Slide]

And if the officers were here, by the way, if I had included officers in the slide, they would be right there, right in between the controls and the administrators.

As part of the study, we have focused a lot of attention on the way in which people eliminate dioxin from their bodies, because that's an important consideration when trying to estimate the initial dose. For that purpose, in 1987 we identified all Ranch Handers that had body burdens above 10 parts per trillion, which is by the way the 98th percentile of the control group. We identified about 500; there were about 500 in that category to be selected for repeated measurement for the rest of study to observe their full body elimination of dioxin.

So that led to estimates of the elimination rate which were at the beginning, and which we realized right away were hopelessly biased because the response variable

that we're measuring was based on a truncated dataset, that we were selecting individuals because they were high. That's a standard environment for an artifact in statistics called regression to the mean.

Well, during that period, the 1990s, we devised a way to force the SAS PROC GLM to produce unbiased estimates even in the presence of a biasing effect of selecting individuals for being high. This is the same effect you see when you give students a test and you select individuals that score high on the test and then you test them again a few weeks later, you'll be just a little bit disappointed. You'll find that they have regressed towards the mean. That's an effect that you see whenever you select individuals for being high or low on a continuous variable.

DR. CAMACHO: Isn't there something about a fallacy of regression involved in this? There's going to be a little football around the line, it's spread, it's going to look like a football.

DR. MICHALEK: Exactly.

DR. CAMACHO: If you do it later, it always looks like the bottom came up and the top came down.

DR. MICHALEK: That's regression to the mean.

DR. CAMACHO: That's what you're referring to now.

DR. MICHALEK: That's right.

DR. CAMACHO: All right.

DR. STOTO: But they're only looking at the top half of it, so. You see the top coming down but not the bottom going up.

DR. MICHALEK: This algorithm was published several times during the period, and it's used in all of our recent papers on estimating the half life of dioxin in the Ranch Hand cohort.

The latest estimate is that the half life of dioxin in Ranch Hand veterans is about 7.6 years at a 95 percent confidence interval.

[Slide]

Here's a picture of the log units, the dioxin level is decreasing in the right chamber over the four repeated measurements of the -- roughly 300 individuals have repeated measurements across all four study cycles.

Remember, we took the first measure in '87, then we went back to the freezers and extracted serum from our freezers and measured the serum that was collected in 1982. And we continued that up to 1997. Our most recently published paper concluded that we should not continue the pharmacokinetic study because so many individuals were getting into background levels, the variance of the estimate was actually increasing rather than decreasing with increased repeated measures. So there was no statistical gain to continuing that study.

[Slide]

Here you see the increased body fat over time in this cohort that was in the pharmacokinetic study, our study.

There is a strong relationship between the body fat and the elimination rate. Heavier individuals hold onto their dioxin longer. They have a smaller elimination rate. And here you see the elimination rate plotted against the body fat measured in 1982, and we see a downward trend.

That's an important consideration in all of our statistical analyses. In all of our reports we adjust for body fat for this reason, because we're trying to accommodate the known variation in the elimination rate with body fat.

DR. HARRISON: that doesn't look very -- What's the R-value for that thing, Joel?

DR. MICHALEK: It doesn't look very pretty, does it? But remember, this is an uncontrolled study and that's the way it is.

DR. HARRISON: Wait a minute, though. Where did that line come from?

DR. MICHALEK: The line is a least squares line from the analysis to produce the elimination rate.

DR. STOTO: It actually looks quite high, if you would drop out that one point with the negative elimination rate.

DR. MICHALEK: Now this guy we can talk about. Why does he have a negative elimination rate? That's because his dioxin level went up. And the reason it went up is he went to work for a utility company in Kentucky between 1992 and 1997 and he was handling transformers and electrical equipment. We think that's where he got his dioxin from.

Remember, these are free-living individuals, they're all exposed to dioxin in the United States, just like all of us in their job and in their leisure activities. So what we got is an exposure that took place many years ago, and overlaid on that we have some noise; from exposures that were experienced here in the U.S.

DR. HARRISON: That R is like .35, right? That R is like .4, right?

DR. MICHALEK: Possibly. Yes, I can find out.

DR. STOTO: But if you took out that guy in Kentucky, it would be substantially higher than that.

DR. MICHALEK: Probably less, yes.

DR. GOUGH: The R would be higher, or the slope of the line would be more acute?

DR. MICHALEK: I don't expect that -- that's not the influential point on the slope. This one's influential, but that one is probably not.

DR. HARRISON: It almost looks like something is tethering it around that 10 percent mark.

DR. MICHALEK: We called him up to talk to him; his levels were coming down nicely. They are like 80 parts per trillion, 60, 50, 90. "Where were you? What did you do between 1992 and --" "Oh, yeah, I got this job."

So things happen, and that's just a reminder that these are not animals, these are people, and we can't control what they do.

DR. HARRISON: How was that assessed?

DR. MICHALEK: How was what?

DR. HARRISON: How was the percent body fat assessed? I forget.

DR. MICHALEK: That's simply the body mass index times -- a later function of the body mass index, weight over height squared, in metric units.

DR. HARRISON: Has that assessment throughout that study ever been -- you know, I went to the Bills-Chargers football game last week, and I saw literally a ton of individuals who had body mass indexes in the obese range, but who literally had no body fat. You know, those are highly trained athletes, highly muscled athletes.

At the other end of the spectrum, there's something referred to as the sarcopenic female. That's a woman who has a more or less normal body weight because she doesn't eat much, but who has more than normal body fat because she doesn't exercise much. So she has a normal body weight but she has a high percent body fat.

DR. MICHALEK: The body fat measurement was discussed many times through the study. The current method is being used for a lot of reasons. The gold standard, I believe, is the immersion method in a tank of water?

DR. HARRISON: Sure, but you've got bioconductance, which is a pretty convenient way and is reasonably close to -- I just wonder if you have -- for instance, if this were a prison population, you'd be overestimating body fat because those guys have nothing to do but work out all day. I just wonder.

DR. MICHALEK: Since the study began, there are new and better ways to measure body fat. In fact, we have a clinical study of insulin sensitivity happening right now in Little Rock, Arkansas. There they're using something called a bod-pod, which is a chamber in which you sit and then you displace air. And of course that has its own limitations, but that might be better -- I don't know; I haven't seen any literature on that.

But there are probably higher technology ways of measuring body fat today that didn't exist in 1976, which is when the original concept of body fat, where weight over height squared was specified.

DR. HARRISON: I just wonder, Joel, if you could even say that in this population there was greater variability or less variability using the BMI, that would at least allow you to comment on the scatter that you see.

DR. MICHALEK: I can tell you the BMI is widely used in our studies; it was used in the Vietnam Experience Study, it was used in the NIOSH study.

DR. HARRISON: Well, the BMI is -- that's the standard.

DR. MICHALEK: I know. It's the standard, and not only that, it's noninvasive. I'm not arguing that there may be better technological ways to measure body fat, and those should be considered for the next physical.

DR. HARRISON: Heck, you can go to Brookstone and get one of these little things, you hold it in your hand like that, and it does bioconductance.

DR. STOTO: Well, two things. One is that if the BMI is an imperfect measure, presumably the R-square would go up if you had a better measure, in this discussion here.

DR. HARRISON: Right.

DR. STOTO: And I guess the second thing -- we should think about this tomorrow. You know, is it worth trying to do some of these other, more precise measures?

DR. MICHALEK: At this stage of the game.

DR. STOTO: Yes. I don't know what the answer is, but I think it's worth talking about.

DR. HARRISON: My question was had it ever been done, and your answer is no, you've not ever correlated the BMI in your study population with any other more precise measurement of body fat.

DR. MICHALEK: No, we have not.

DR. GOUGH: Didn't you do some immersion studies on a --?

DR. MICHALEK: We thought about it, but we gave that up. Because it's not a very pleasant experience for an older gentleman to be put into a tank and told to exhale and stay completely exhaled until some technician says, "Okay, you can breathe now."

DR. GOUGH: Or you sink to the bottom.

DR. MICHALEK: It's not fun.

Oh, you just thought about it.

LTC BURNHAM: Our oldest subject is 80 years old --.

DR. MICHALEK: I did it once. I would not like to do it again.

Yes, so that's an issue. Body fat is an issue.

MS. del JUNCO; 92.

[Slide]

DR. MICHALEK: Here's a comparison with some other studies on half life. Here is the Ranch Hand study 57.6 years. There was a study of individual adults in Italy in the Seveso accident done by CDC, and another study of -- you saw the previous slide of those observing the industrial workers; these are smaller studies based on paired measurements and our study is based on up to four measurements per subject. Roughly the same ballpark, which gives us confidence that we're working in the right arena.

[Slide]

Here are some Ranch Handers at a museum in Hurlburt Field, a Ranch Hand aircraft.

Do you want to say a word about that airplane? Do you happen to know anything about this particular aircraft?

MAJ SPEY: It was not a spray airplane, sir; when it was moved to the airpark, why we convinced them to put spray booms on it just for fun. It was an airlift airplane in Vietnam.

LTC BURNHAM: Is the one over at Lackland originally a spray air? There's one outside the gate at --

MAJ SPEY: I'm not sure.

DR. MICHALEK: Here you see a representation of the spread of conditions that were being reported by Vietnam veterans; and those form the structure for our study.

I'm now going to run you through, show you an overview of findings, and this will be layered. In other words, today I'm going to show you an arm's length view of everything, and then we're going to focus down to some particular areas such as diabetes and peripheral neuropathy.

We have produced about 20,000 pages of reports, almost all of which have been written by Science Applications International Corporation, by means of a study design and statistical analysis plan, which is based on these statistical models.

We have four approaches to analyzing data in the study. In the first approach, we don't use dioxin measurements at all. We just compare all Ranch Handers with all controls. And then within these three occupational categories, we compare Ranch Hand officer with control officers, and so on. It's about a one.

Separately, the next three models use the dioxin body burden. In those Ranch Handers that have high levels today, that means more than 10, we extrapolate back to Vietnam, and ask whether the initial extrapolated dose is related to current health. That's called the

Initial Dioxin Analysis, or Model 2.

Separately, we categorize individuals into four bins; with controls, and then we take the Ranch Handers and break them up into three parts: Those that have background levels today and then those that are above background where we break them out to low and high. And we compare each of those three Ranch Hand strata with the controls. That's called our Dioxin Category Analysis. That's the way you'll see it primarily in all our published papers.

And then finally we ask: Is there a connection between today's dioxin body burden and your health? No matter how much you had in Vietnam or where you got it from, is there any connection at all between today's dioxin body burden and health? And that's our Model 4.

These are the four models that were used in our 1997 report, which is on the web page. In our 1992 report, we used six models, where we added two more dioxin level analysis at the bottom here that I'll talk about later.

[Slide]

Here are some sample sizes of the numbers of people that came to a physical exam that were in the strata used in the first model. See number of officers, enlisted flyers and ground.

370 Ranch Handers came to the physical, and 1251 controls. We have about an equal number of enlisted ground as we did officers.

[Slide]

Here's dioxin category numbers, and here are those four bins I was telling you about. Here are the comparisons, and in the comparison group we eliminated the one percent or so of the comparisons that had greater than 10. Because some of those, we believe, received high levels here in the United States, by means of their occupation.

So because of our philosophy of wanting to study exposures that occurred during the war, we wanted one to focus on war-related exposures, and that's why they excluded the top 1 percent of our comparison group.

And by the way, even if you put those people in, the analysis results generally don't change.

Here you see the three categories in the Ranch Hand analysis. The low and high categories were defined by their initial dose in Vietnam, the median level that 94 parts per trillion. That's the split that broke this group up into parts of roughly equal size.

The analysis drill is to compare each of these, and their health, with the comparisons.

[Slide]

This is a thumbnail sketch of what we saw, not just in the last report, but in all available data. In the area of general health, I guess the finding that I remember most is that we see a significant, adverse relation between reported health and dioxin body burden. Reported health on a scale of excellent-good-fair-poor. We see an increased risk of reporting fair-poor health in the high dioxin-exposed category, in our dioxin category analysis.

That was a point of discussion at our previous meeting, and I have some slides on that. In October of last year, why are we seeing this and why did we see it in previous reports? What does this mean? What does the general assessment of health mean?

Since then we have looked and we have found that that particular assessment is significantly related to diabetic status. Meaning that, at least part of what they're recording is their diabetes, which is interesting. Because that thread of thought will prevail through many of the findings in the study.

We see so far no relationship, or no significant relationship between any measure of exposure and cancer. However, that's certainly an issue we look at very carefully. That's been looked at of course in all of our reports, but it's recently published in the American Journal of Epidemiology, 1999.

The latest report from SAIC, just recently released in January of this year, we see a 6 percent increase in cancer in the whole group; which is of course not significant. About 16 percent of all Ranch Handers and comparisons have one or more tumors, at this point.

So we have very good statistical power to detect relative risk of 2. We have no statistical power to detect a relative risk of 1.06.

In neurology, because of our work with the National Institutes of Health and the National Institute of Dental Research, we have collaborated with a physician at the University of Michigan to measure peripheral neuropathy in the most thorough way that we have ever done, and we have found a significant and adverse relationship be peripheral neuropathy and dioxin body burden, and that is in submission to a journal, and I'll tell you more about that in a separate talk on that.

In psychology we're seeing generally no relationship between any measure of exposure and any measure of psychological health except -- that means the MMPI, the SCL90R, and all the measures we've given to the study. If you look at our web page and click on our reports, you can look at the cite chapter and you will see all the different instruments we've given since the beginning.

However in 1982 we gave, in addition to questions about anxiety and depression, we administered the Wexsler memory scale and the Wechsler adult intelligence scale, and the Wechsler reading achievement test, the RAT. Those results are recently now analyzed and are in submission to a journal. We see a significant and adverse relationship between short term memory and dioxin body burden that we had not seen before, because only now have we gone back to analyze data in 1982 cognitive function.

That data is interesting because it's consistent with results seen in babies of women who were exposed to PCBs in studies done in Amsterdam, in Holland. And those are recently published.

MS. GOVAN: Joel, when you're identifying positive versus negative findings, are the findings that are positive mean that it had to fit that monotonic, linear relationship from low to high? And if it's negative, there could have been an association, but it would fit that linear pattern?

DR. MICHALEK: Certainly the first thing is true; If it's a positive, that means there's a significant adverse relationship there, a positive trend with dioxin body burden. If it's negative that means we're unable to find any pattern there that made any sense.

I have a separate talk on cancer where I actually show you the data. What happened on cancer was, that we see an increased risk of cancer in the low group but not the high. In fact, we saw a decreased risk in the high group. Difficult to interpret. So we interpret that as negative.

DR. CAMACHO: Are the numbers in the cells in all of this --

DR. MICHALEK: 300, roughly. We have small numbers. Certainly this study has no ability to study rare diseases such as a particular sarcoma. It has good -- we're getting into another talk.

This physical power to study all cancers combined -- in the area of gastrointestinal, we look at history of liver disease -- and by the way, all diseases are verified by medical record review, 100 percent. So we looked at liver disease, we looked at liver enzymes and liver function. And we see a consistent and adverse relationship between certain liver enzymes such as UDT and dioxin body burden, but no evidence of a relation between liver disease and dioxin body burden; and that's currently in submission to a journal. That's also been described in our reports.

DR. HARRISON: What about gastrointestinal functioning?

DR. MICHALEK: In what regard.

DR. HARRISON: What about, let's say the incidence of patients taking medication used to treat peptic ulcer disease, taking medication used to treat gastric motility problems?

DR. MICHALEK: Have not studied those endpoints. We've studied ulcers, we have not studied medication as an endpoint.

DR. HARRISON: For instance, patients with diabetes will at some point, or can at some point have difficulty with gastric emptying. So you'd expect to see, if you had a big enough population, you might expect to see some evidence of that.

DR. MICHALEK: Well, that idea is certainly captured in the minutes, and we'll--

DR. HARRISON: And it would go along with your positive peripheral neuropathy because these are all neuropathic problems, and the more you tie those together, Joel, the tighter you make the story.

DR. MICHALEK: The picture.

Cardiovascular, we're seeing an overall 25 percent increase in cardiovascular disease in the Ranch Hand group. Again, all verified by medical record review. That's separate from cardiovascular mortality.

We are talking about here the health effects we see in the veterans who've come to Scripps Clinic.

However the patterns after that are not completely clear. We see --

DR. GOUGH: Joel, did you say 35 percent?

DR. MICHALEK: 25 percent.

Yes?

MS. GOVAN: Could you describe a little bit about -- that's such a big, broad brush.

DR. MICHALEK: I know. I have a separate talk on that, too.

DR. HARRISON: Ma'am, this is basically an overview to try and get the committee up to speed on what has happened overall.

DR. MICHALEK: There is a wide range of ICD codes that cover that definition, and I'll have to address that separately.

[Slide]

Hematology, we're seeing a significant and adverse -- I couldn't call it adverse, because I believe people know what's adverse here. But we're seeing changes in platelet count and mean volumes with dioxin body burden; and that's in submission to Archives of Environmental Health. The meaning of that is unclear or unknown.

In endocrinology, of course we're seeing the significant -- we have a lot to say about diabetes today. We're seeing a relationship between diabetes and dioxin.

Immunology, published in the American Journal of Epidemiology, 1999, we see no detectable adverse relation between any measure of exposure and immune function.

In pulmonary we primarily no relation except among officers we saw an adverse relation between -- bronchial obstruction. There was a finding in our 1997 report, and that's the reason for the plus-minus.

In dermatology we've seen, as I said, no evidence of chloracne with the caveats that we stated.

And in renal, no relationship between any regular exposure and renal function early in disease. Not expected, either, in renal.

[Slide]

Here are some numbers showing you what the demographics were in 1995 after Cycle 4, the fourth physical, of what the ages were, all of the categories of our dioxin exposure

index. You see the individuals in the high category are slightly younger, and the individuals in the background are slightly older than controls. That reflects the fact that most of the individuals in the high category were enlisted, and most of the individuals in the background category were officers, and officers are generally older.

[Slide]

Here you see that the pattern in body fat parallels the pattern I just described by occupation.

Here you see the percentages by military occupation in the high category are 2 percent for officers, whereas in the background category, 61 percent are officers. Which is an important adjustment in our analysis, because officers are generally college-educated and enlisted are not. So we have to be careful to make these variables part of our statistical modeling.

[Slide]

Here you see what diabetes looked like in 1995, which was a pattern of increased relative risk from background, low/high, .7, 1.3, 1.5, and that 1.5 was significant, and there's a lot to say about that during our meetings today.

Here's what it looked like in 1998, the same increase, the prevalences are increased. Back up here you see a 20 percent diabetic in the high category and here 23, almost 24 percent diabetic in the high category.

[Slide]

Here's what cancer looked like in the study. This is what I was telling Debbie about just a few minutes ago. We see a pattern of increased risk here, but not here. After adjustment for many covariates. This is all cancers.

Heart disease, we see a pattern here which is not very exciting statistically. We see a relative risk of 1.0 and not a category; that's what I meant, the cardiovascular findings are a puzzle. We see an increased risk overall, in all Ranch Hand groups. We see this one, we do a dioxin category, but we see an increased risk of, evidence of prior myocardial infarction when we look at the initial dioxin body burden in Vietnam.

Yes.

DR. HARRISON: Dr. Sills knows more about this than I do. But patients with diabetes don't have clinical heart attacks, but they do have subclinical heart attacks, and they have more of them. They're smaller.

Part of the mechanistic explanation for that is that they have more atherosclerosis, they have more partial obstruction, and so they produce more bypasses on their own, so that when they do finally knock one off, they knock of a smaller, more localized piece and frequently just don't have chest pain and don't have any symptoms and go about their business.

So depending on what -- see, if what you're calling heart disease is the medical record that this patient had a myocardial infarction, then that should well be different from, if you did EKGs on everyone and found this puzzling observation, that a lot more of the high group had abnormal EKGs.

DR. MICHALEK: Which we do find.

DR. HARRISON: Do you agree, Dr. Sills?

DR. STILLS: I just want to point out I'm a veterinary pathologist.

But I agree with what you say.

DR. GOUGH: Joel, before we leave the slide, if you do the comparison between Ranch Hands and comparisons, is there a difference? Non-stratified.

DR. MICHALEK: Yes. We see a 25 percent increase.

DR. GOUGH: Is it statistical significant? I mean, those numbers aren't.

DR. MICHALEK: I wouldn't be surprised. You know why? Because the prevalence is 65 percent. 65 percent of both groups had some condition which counted towards our definition; so we have very high prevalence and we have very power. We probably did have significance or borderline significance on that 25 percent.

DR. GOUGH: Well, this is a strange dose response.

DR. MICHALEK: It is. It certainly is. But there's a lot of complications here, as mentioned by Dr. Harrison. And there's --

DR. GOUGH: But see, I ignore Bob's complications. I can never understand them.

DR. MICHALEK: There could be a problem with our definition

DR. HARRISON: Let the record show that the Chair has been dis'd.

DR. MICHALEK: There's literature out there to suggest that dioxin destroys vascular tissue, and that we may be just looking at the date incorrectly. There's a lot of ways to look at this data, and that's why we're having this meeting.

DR. GOUGH: But to follow up on something Bob said, you have EKGs on everybody, right?

DR. MICHALEK: Yes, we do, and that's one of our endpoints.

DR. GOUGH: So that's factored into this?

DR. MICHALEK: No, this is a definition by -- by ICD code, there was a definition of heart disease --

DR. GOUGH: Oh, okay.

LTC BURNHAM: Is carotid thickness in here, too?

DR. MICHALEK: No, carotid thickness is a separate analysis which is being done by Billy.

DR. GOUGH: Okay.

DR. MICHALEK: Not part of the SAIC report.

[Photo]

Now we're going to talk about mortality. This is a moment to the Ranch Hand killed in action at Hurlburt Field. We did the standard breakout by unaligned cause of death. These are the same categories used in many other studies.

Overall, through 1993, we see -- relatively we see nothing. We see an observed 118 deaths in the Ranch Hand group after Vietnam, and expected 120, both risks less than one.

However, when we look by cause of death, here we see a finding, we first noticed in 1988 increased risk of death from cardiovascular disease in the enlisted ground crew. And that has persisted ever since. And all the other areas we see no evidence of an effect of any note; especially in cancer the relative risk is .9.

Remember, what we're talking about in mortality is a comparison between the observed and the expected number of deaths in Ranch Hands as compared to the death rates in the 19,000 in our control population. We do not have dioxin levels on 19,000 controls. We are not able to adjust here for dioxin body burden. We are only able to adjust for date of birth, race, and military occupation. That's all we've got in the way of covariates.

That causes us to be concerned about this digestive death relative risk of 1.7, which is significant. We know that many of these deaths were due to alcohol abuse. We're unable to adjust for alcohol consumption in these mortality analyses. We're also unable to adjust in the cardiovascular area, for example, for cardiovascular disease in the family, which is a risk factor. We're unable to adjust for smoking, which is a risk factor. We're unable to adjust for any of the standard risk factors that we're able to do when we look at data coming out of Scripps Clinic.

Bill Grubbs and SAIC have access to all the covariates; we do not have those covariates with mortality.

Yes.

DR. HARRISON: What about the increased risk in infection?

DR. MICHALEK: Those are small numbers. Two individuals here in 1.3, I'll find out what those were and tell you what they were; I have to look at the records.

DR. HARRISON: I'm sorry; I see. Okay.

DR. MICHALEK: Small numbers.

DR. HARRISON: I agree.

DR. MICHALEK: The other arm of the study is reproductive outcomes.

We have identified all children, live births, 8,100 children. We have identified and verified their lineage -- their existence, their lineage and their health up to the age of 18, by means of medical record retrieval and review.

We have identified all 10,000 conceptions that were produced by these men over their entire life, by medical record review of the records of the mother, primarily.

Separately, we have measured sperm parameters on the men themselves, and certain gonadotropins such as testosterone and FSH and LSH.

And here are the endpoints we studied. I think I have a slide. We primarily see no result when we ask whether there's a relation between any of these conditions and any measure of exposure. There's a few exceptions.

In the area of hormones. In testosterone, if you study levels of abnormally high testosterone -- as we did in a published paper in 1996, I believe -- you'll see no relation between abnormally high testosterone or abnormally low testosterone and dioxin body burden. However, if you look at testosterone mean, averages of testosterone, you'll see a significant decrease, a slight decrease which is statistically significant, because we have enormous statistical power when studying averages. And that's published in Epidemiology.

In the area of birth defects, we see no pattern which was considered meaningful or suggestive by CDC. With the exception of spina bifida, we saw in our dioxin exposure analysis zero cases in the control group; zero in the background category of the Ranch Hand group; one in the low group and two in the high group. That pattern of 0 1 2 was declared suggestive by the National Academy of Sciences, and that led to compensation to all Vietnam veterans of spina bifida in their children.

So that is the reproductive finding so far that has been recognized: The pattern of increased risk of spina bifida. We couldn't handle that statistically because the numbers were too small.

[Slide]

And here is a description of the check mark pattern. We have a picture of that, and I'll show you that in a second.

The pattern is represented simply by a trend in the Ranch Hand group, from background-low-high, and yet an overall relative risk of approximately 1.0. What that will be realized as, relative risk of less than one among individuals in the background category, and a relative risk of greater than one in individuals in the high category.

That was first interpreted in 1992, as possibly an artifact of reverse causation. And it's been talked about in the National Academy of Sciences books on Agent Orange in Vietnam veterans.

During the last decade, we have devised a simple misclassification model to explain the pattern.

[Slide]

But things have changed in the last couple of days, and I need to tell you about that. And here's a picture of the histogram again I showed you earlier, and here's a statistical model of the back up. Here's the normal distribution of the Ranch Hand group and the control group, and here's a statistical model, there's the rent control distribution, there's the Ranch Hand distribution today, and there was the Ranch Hand distribution as we think it should have been many years ago, before they lost their body burden, before it decreased.

Here's the picture we see today in diabetes. That is, up until about 3 o'clock yesterday.

Up until 3 o'clock yesterday -- this is what I want Mike Stoto to hear -- we have been analyzing diabetes with logistic regression adjusted for body fat, age, family history, and other covariates. And we consistently see this pattern.

What has happened in the interim is that we have written software to match, one-to-one, Ranch Handers to Comparisons on body fat when they were in Vietnam to within three percent. Family history of diabetes in the parents, brother or sister, perfectly. And date of birth -- nearly perfectly. And race. And military occupation.

DR. STOTO: This is not the regular --

DR. MICHALEK: No, this is super-matching.

DR. STOTO: -- standard match; this is new matching.

DR. HARRISON: Super matching.

DR. MICHALEK: I'm setting you up. Are you ready?

DR. HARRISON: Did you say super match or super magic?

DR. MICHALEK: This is super matching. This is maximal. And what's great about it is that body fat was measured before they were even exposed.

There's no issue here about reverse causation, about dioxin body burden changing your body fat. That body fat was measured in Vietnam.

We did matched pair analysis. The new relative risk in the background category is one. The new relative risk in that median is higher; it's about 1.2. And the highest, 1.5. And overall the relative risk is 1.2, significant.

DR. STOTO: I'm sorry?

DR. MICHALEK: I'm losing you.

DR. STOTO: Yes.

DR. MICHALEK: What happened was,

first of all, the check mark pattern went away, when we do a matched analysis, highly matched, the way I said.

You see this? We don't see this anymore. This is what we've been seeing for the last --

DR. STOTO: What are the three graphs corresponding to?

DR. MICHALEK: See, I'm so excited, I can't even tell you.

(Laughter)

DR. HARRISON: Are you saying that that right graph is the supermatched groups?

DR. MICHALEK: Yes. Up until 3 o'clock yesterday, this was a graph showing what we expected to see according to the statistical model that these slides were supposed to talk about. What I'm saying is, we don't need that statistical model anymore; dump it. I'm telling you that with this supermatching that I just described, this is what we see in the real data.

We don't see this anymore. This is what we saw using old-fashioned logistic regression. This is what we see when we do very careful matching.

That means that this diabetes as a disease is very sensitive to these factors, and that your body fat when you're young is very predictive.

COL MARDEN: Which way?

DR. MICHALEK: Adversely related.

COL MARDEN: So more body fat means more absorption.

DR. MICHALEK: Higher fat individuals have an increased risk of diabetes.

Yes.

COL MARDEN: So more body fat, more absorption and body burden, more diabetes.

DR. MICHALEK: yes.

DR. GOUGH: When you say body fat, you just mean height and weight, right?

DR. MICHALEK: Yes, it's basically BMI.

Now of course this is brand new. It isn't even out of the -- so everything I'm telling you today is going to be checked out over the next several weeks. Yes.

DR. GRUBBS: Joel, the additional adjustment factors here, to summarize, are?

DR. MICHALEK: Family history, body fat in Vietnam.

DR. GRUBBS: Okay, body fat before exposure.

DR. STOTO: Let me see if I can restate what I understood you to say.

When you control in this new improved way for the known risk factors for diabetes, the relationship between exposure to dioxin and diabetes is strong and the check mark problem goes away.

DR. MICHALEK: Exactly.

DR. STOTO: Okay. That is pretty important.

(Simultaneous conversation)

DR. MICHALEK: In other words, the pattern we see becomes sharper. The picture comes clearer.

DR. GOUGH: What do those symbols

above the second and third box mean? Exposure, or respective --

DR. MICHALEK: This is that statistical model I'm talking about. This is the distance separating the distributions. We back up one.

You see this distribution, Mahalanova's distance units. This is the controls today and that's the Ranch Handers today. The distance in Mahalanova's distance units is the difference of the means over the standard deviation. That's about 1.5 today.

If you imagine what the Ranch Handers looked like years ago, they were probably out here. Now the Mahalanova's distance is 2.5.

I'm able to statistically model this pattern in terms of that single parameter called Mahalanova's distance. I can make the pattern go away and I can make it come back. I can make it go away by making a bigger distance, and I can make it come back by making it a smaller distance.

Here is the observed pattern and here is the expected pattern. This is what we see today and this is what is predicted by the model. And I can make the model go away by moving those distributions apart.

But I can dump all this now. Forget it, we don't need it anymore. The purpose of this was to think, well, maybe this check mark pattern was due to misclassification. You know, we're being misled. The day is fuzzy and they're far apart and they're closer together now than they used to be, and our statistics are all screwed up because of it. Dump it. We just did this matching, we don't need this anymore.

DR. GOUGH: Well, the other thing that is really striking to me is that, I thought in the past, when comparing Ranch Hands versus Comparisons, that the incidence of diabetes between the two groups is essentially the same.

DR. MICHALEK: That's not true anymore.

DR. GOUGH: Well, that's what I asked about.

DR. MICHALEK: Because that's unadjusted. Yes, unadjusted, the overall is about 17 percent in both groups. But that's unadjusted.

DR. GOUGH: But when you adjust on the basis of family history, obesity and race?

DR. MICHALEK: Obesity in Vietnam, race, family history, and military occupation, I don't have the percentages. But now the relative risk is 1.2. There's a 20 percent in the Ranch Hand group, and the confidence interval does not include 1.0.

DR. GOUGH: This is distressing to me, but a clearer picture is emerging, for sure.

DR. MICHALEK: Well, we're going to have a meeting in December. I'll have a separate talk on this in December.

DR. CAMACHO: So in plain English--

DR. MICHALEK: In plain English, there's an increased risk of diabetes --

DR. CAMACHO: If you have two guys in Vietnam, both of them enlisted and one's chubby and one's thin. They both get the same exposure. The guy who's chubby has a higher--

DR. MICHALEK: Higher risk of diabetes.

DR. CAMACHO: -- risk of diabetes. Okay.

DR. MICHALEK: As he was exposed, and another chubby person in Vietnam who didn't get exposed.

DR. CAMACHO: Who didn't get exposed.

DR. STOTO: I think that supports original check mark theory, by the way.

DR. MICHALEK: Yes, it does, by the way. It supports everything that's happened in the last few months. That there is a relationship between diabetes and dioxin.

DR. HARRISON: How many people in the Ranch Hand group with diabetes? In other words--

DR. MICHALEK: About 16 percent out of 1000 --

DR. HARRISON: So you're saying there's 150 to 200. So what you have is 150 to 200 on this side, and then you picked 150 to 200 exact matches on this side.

DR. MICHALEK: No, no, we didn't match them. You're talking case control. We matched cohort. We took every Ranch Hander, whether they had diabetes or not, and we matched them perfectly to a control.

DR. HARRISON: Oh, okay.

DR. MICHALEK: And then we looked at differences on diabetes. And we stratified by dioxin body burden, and we see this.

COL MARDEN: And this check mark was in over 50 different analyses.

DR. MICHALEK: Oh, yes. We saw it in body fat --

COL MARDEN: So it does make you think that it was the statistical analysis rather than something specific to diabetes.

DR. HARRISON: It's always statistical analysis.

(Laughter)

COL MARDEN: This is true.

DR. MICHALEK: So we're going to go back, we're going to check to see if we can make some other check mark patterns go away with this careful matching.

MS. del JUNCO: Joel, and the results were significant for both groups?

DR. MICHALEK: Say that again, please?

MS. del JUNCO: The results were significant, the confidence intervals were significant for both groups for diabetes?

DR. MICHALEK: The results. Yes, the relative risk is significantly increased overall, and is significantly increased in the high category.

[Slide]

This is an overall, thumbnail sketch of the whole study. We've talked about all these things. And we've made a lot of reports, and they're all available on our web page.

This is just a quick overview of all the papers we've published. I know I'm running out of time, so what should we do? We were supposed to stop at ten.

And Jay hasn't done his slides yet.

I can stop here.

DR. STOTO: Can I just report that on this, we talked about whether the heart disease would be significant if you lumped all the Ranch Hands together? I think that the answer is yes.

DR. MICHALEK: I think we needed to check that.

DR. STOTO: I just tried to do -- I think the answer is yes.

DR. MICHALEK: It is?

DR. STOTO: Yes. Not adjusting for anything else, obviously.

DR. HARRISON: Well, we started a little late, so why don't we plan to go until 10:15, and then we'll take our break. Is that enough --?

DR. MICHALEK: In other words, I should finish up, and --?

DR. MINER: Yes. Go ahead, Joel.

DR. STILLS: Can I ask one quick question? In terms of the neuropathy and the cardiovascular disease, are you only seeing that in your group that is significant diabetes?

DR. MICHALEK: Yes, I have a talk on that, too. Of course peripheral neuropathy is highly related to diabetes. In fact, the relative risk of having peripheral neuropathy is about 30. Diabetics have about 30 times the risk of peripheral neuropathy of non-diabetics.

So in our analysis of that variable we had to be obviously very careful about diabetes. Are we seeing simply another reflection of diabetes or not? The end analysis was done with diabetes in the dataset, with diabetics in the cohort included, and then as a covariate, and it was also done with diabetics excluded. And we still saw a significant increase in risk of peripheral neuropathy.

But when we went back and looked at the medical records of every case of individuals that were diagnosed as having peripheral neuropathy, there was always some mention in the record of glucose. Even though they aren't called diabetic yet. It's interesting that the physicians wrote, "something to do with glucose or insulin" in their record.

DR. HARRISON: Well, that may be a self-fulfilling prophecy. I mean, if I see someone with peripheral neuropathy, I'm going to write in my notes that I have to rule out diabetes. So if you're just scanning, that's -- phhh.

DR. MICHALEK: Yes. But the point is well made that the two outcomes are highly related and they were addressed in our analysis.

So we have written many different papers, and these are the areas that we've published:

[Slide]

Statistical methodology, health endpoints, pharmacokinetics and dioxin levels. And many of those are published, of course, and some are in submission and some are out right now.

I want to emphasize here, something we failed to emphasize when we talked to GAO. GAO said in their report we didn't start publishing until 1990. That's not true. We actually launched our research immediately; and this first paper, published in 1980 -- actually, as you know, you write these things; they take years to write and get published. We began that work in 1977.

So we had papers published initially in Statistical Methodology because we were told that we would be working on a large cohort study using matched analysis; and the primary emphasis at that time was survival analysis. So were studying linear rank procedures and the Cox model and logistic regression and things like that during the period in the '80s, before we published our first health paper in JAMA in 1990.

And remember that the JAMA papers published in 1990 actually began in 1985. When we started to write those JAMA papers, we initially wrote them to include data from the Cycle 2 physical. But then working on the papers, the Cycle 3 data came. And so we updated the article to include only Cycle 3, or 1987 data.

So the activity of publishing began in the middle '80s; it didn't begin in 1990. That's all this is about. More papers on hypothesis testing, discriminative analysis, reliability theory -- these were all coauthored with visiting faculty that were working with us at the time. Published in Biometrica, Biometrics, Statistics in Medicine and other journals like that.

[Slide]

And we've continued up to the present day; we're still writing methodology papers in statistics. We have a paper in progress; we had a paper on calculating P-value that sounds fairly -- why are we doing that? Well, there was always a disconnect between the P-value and the confidence interval and the SMR, which we fixed, and published in the American Journal of Epidemiology in 1998. And recently writing papers on estimating new parameters in epidemiology such as lethality, and we'll talk about that later.

[Slide]

Now the first health paper was published in JAMA in 1990. As I said, the work actually began in '85.

And diabetes was first mentioned, the first published mention of it occurred because of the talk in 1991 or '92 in Helsinki, Finland at the International Dioxin Conference. And that was published in their proceedings, in work on halogen compounds.

Subsequent to that we had papers published in epidemiology and gonadotropins and diabetes. This is the primary diabetes paper which led to a talk earlier this year to the National Academy of Sciences, which I'll tell you about.

And this is the paper on chloracne that I already mentioned, and we have an interesting paper, I'm showing a strong relation between insulin and sex hormone b_globulin and dioxin in the Journal of Endocrinology and Metabolism.

A paper on cancer and immunology,

1999, American Journal of Epidemiology. And another paper on diabetes in Epidemiology showing a relationship between dioxin body burden and dioxin in our control group, which was reported by -- first authored by Matt Longnecker and that was reported to the National Academy of Sciences this year.

And papers on mortality and a letter to the editor on the possibility of differential binding of dioxin to lipids in serum in 1998.

[Slide]

In Reproductive Outcomes, all of our data on reproductive outcomes has been published in one form or another, except for fertility, and that's an article that we -- we went to lunch with Debbie del Junco, and she is with us here today.

Primary birth defects, a paper published in 1995. The work on it actually began in 1984 when we began to verify all health outcomes, all birth defects among all children followed by these men, by medical record review.

Sex of children is an issue. In the Seveso cohort it is shown that children born to families who, for whom the mother and father have experienced high dioxin levels and were all girls. So we repeated the analysis in our data and found no relation to the sex of the children and their father's dioxin body burden.

Here's the paper on testosterone (inaudible) published in 1997.

[Slide]

In Pharmacokinetics we had a number of papers published on the cohort with repeated dioxin body burdens and half life, appearing

primarily in the Journal of Toxicology and Environmental Health. The very first one appearing in 1989. And the very latest in 1999, and that's a statistic I quoted earlier, half life of 7.6 years.

[Slide]

Dioxin levels, the very first results from our pilot study at the Red Cross clinics was published in MMR, WR in 1988. And subsequent to that we have our paper on the skin exposure by questionnaire to the enlisted data I already showed you, showing a relation between on the job exposure in Vietnam and today's current dioxin body burden, and the reliability data I showed you was published in 1996; and we have a paper on the comparison group showing the data, the dioxin body burden comparisons published in 1998, and we have -- in the year 2000, which has just recently been accepted, we have shown a significant decrease in the dioxin body burdens in the control group with time, which parallels a decrease seen in cohorts in Germany and other parts of Europe, that cohort body burdens are decreasing; the speculation is that that's due to regulation of industry.

[Slide]

DR. CAMACHO: I think we're missing a page.

DR. MICHALEK: Oh.

DR. STOTO: I think that page 11 of the handout is missing.

DR. MICHALEK: Oh, I'm sorry. Is a page missing?

DR. CAMACHO: I believe so.

DR. MICHALEK: Okay, I can fix that. We have the originals here.

DR. MICHALEK: In submission, this is the paper I was telling you about, into the neurotoxicology . We have some very dull papers in psychology that show absolutely no relation between the MMPI and dioxin, which would be very difficult to publish; but this is in submission to the Journal of Consulting Clinical Psychology. It's been with them now about a year, but before that it was submitted to other journals and bounced immediately, or said 'rejected' -- so this may -- we may never get this published. That's our paper showing no relation at all between the MMPI and any measure of exposure.

This paper is close, it's been reviewed and sent back to the journal. We've responded to the referees; this is showing the relationship between liver enzymes and dioxin body burden.

Hematology, this one is very, very close. It's been reviewed several times by the Archives of Environmental Health, and we've responded once more to the referees and sent it back. We expect acceptance very soon.

Peripheral neuropathy was submitted to the American Journal of Epidemiology, we got a very glowing letter back telling us what a great paper it was, and they rejected it. So we are responding to the referees right now, and we're going to resubmit to Neural Toxicology. That's the paper showing the relationship between dioxin body burden and peripheral neuropathy.

[Slide]

In this, a meta analysis of -- relating dioxin body burden and diabetes with dioxin body burden and diabetes in the NIOSH cohort, and that's in submission to Epidemiology. That gives the expected result; mainly -- you see a trend in the Ranch Hand group and you see a fairly wimpy trend, so to speak, in the NIOSH group, and it won't go away; it causes an interaction and prevents a meta analysis. It was not a very interesting paper.

And finally there was a paper on dioxin and diet which shows no relation between any measure of exposure to dioxin or any aspect of diet, and it's collected from the diet questionnaire given to our study subjects in 1997, I believe.

Is that when we did the diet questionnaire?

DR. MINER: '92.

[Slide]

DR. MICHALEK: In progress right now we have a measure of the carotid artery wall thickness. It was done by Dr. James Dwyer at the University of California, and he has shared that -- of course he's part of our team and he's been working with Billy Jackson, and we're relating that to dioxin body burden. That's one part of a two-part paper on cardiovascular disease and dioxin body burden.

We're seeing a relationship that is puzzling, between carotid wall thickness and dioxin body burden. We have a paper on medical symptoms. These individuals fill out a checklist of up to 30 symptoms at every physical. "I have aches and pains, I can't sleep, I urinate too much" all kinds of things; and those symptoms have never been described as related to exposure, and we're attempting to write a paper on that with CDC. That paper is in the works.

We have a paper on thyroid function with Dr. Arnold Schechter which is just about to start. And a paper on fertility with Ann Sweeney and Debbie del Junco and Kanazi {ph} University of California-Berkeley, which is just about to start. Although here we realize we have to clean up our datasets.

And there's the paper on check mark pattern which has been blown away by the result I just told you about; and dioxin body burden and elimination which I have a talk on that to give you later.

[Slide]

Days in Vietnam was an issue brought up by our advisory committee last year. Forget dioxin; number of days in the country, did that have anything to do with your health? And we have a statistician working on that problem right now. And we have other papers in progress -- others here that I will not read to you.

[Slide]

We have many reports; all of them are available on our web page. We were audited by the GAO, all of calendar year 2000 -- In 1999, sorry, and they released their report in 2000 with three recommendations: Release all of our data, improve our communication, and improve the advisory committee outreach; and all of those things are being done or have been done.

[Slide]

The data release. We are literally releasing everything that we've got to the public by means of CD-ROMs that we send to the Government Printing Office and by means of our web page. You can download datasets that we used in all of our reports. They're there in two formats: in SAS and in flat files. You can point and click and download those.

There's up to 12 clinical datasets for every physical exam. There's one for general health, dermatology, cancer, heart disease, diabetes, endocrinology. All of those are on our web page, all of the laboratory datasets are there, and everything to do with reproductive outcomes, and all of our mortality datasets.

Now we're just about to release all the data collected in 1985 and by the end of the year, we will release everything collected at baseline in 1982.

[Slide]

Limitations are clear; we know these, we've known them when we wrote the protocol. Cannot establish causality. There was a paper published later by Bross in Biometrics which clearly shows epidemiology studies cannot establish safety and cannot clearly say the ____ derivative is one, we don't have sample size to do that.

We don't have the power for rare conditions. All of these things are there, we've known about them. More recently it has been emphasized to us by the veterans that, "Gosh, why did we use a Vietnam veteran cohort for our control group? I wish we had used a non-deployed control group."

Well, that was the thinking back in '77, '78 and that's what we've got. And why did we use Air Force veterans? They wanted to see an Army study, which makes me regret that CDC stopped the Vietnam Experience study.

In accordance, the veterans are saying, we're asking the wrong question looking at Air Force veterans; why aren't we looking at Army troops? That was the Vietnam Experience study, which was stopped.

[Slide]

A suggestion would be, and this was brought up at a meeting at the VA a few weeks ago, to restart the Vietnam Experience study. After all, they received a physical and questionnaire just like the Ranch Handers did back in 1987, and the records are still there, they're in boxes. The study subjects are all identified and all the data is available to do a final examination.

So that idea is being discussed. Another study that has been sitting there and has not been fully published is the twin study. It's a study of about 4,000 individuals who didn't go to Vietnam; it had a twin brother who did go to Vietnam. And that study is being conducted by Dr. Seth Izin at the VA Hospital in St. Louis.

We have the dataset and we've been tracking their mortality along with the Ranch Handers controls, and we're about to give him a dataset so that he can look at mortality, anyway, comparing his ultimately matched study of twins.

DR. HARRISON: Were they raised together?

DR. MICHALEK: What.

DR. HARRISON: These were twins that were raised together?

DR. MICHALEK: Well, I assume they were raised -- I don't know. They're twins.

DR. GOUGH: They separate out the ones who were raised together from the ones who weren't.

DR. MICHALEK: They were born from the same womb at the same time.

DR. GOUGH: They've been doing this since World War II.

DR. HARRISON: Okay. Okay.

DR. MICHALEK: I can tell you more about that later.

Secondly, there's the idea of constructing a new control group for this study, which is certainly a possibility. For example, if there was another large cohort study out there with diabetes for example, we could pass -- and that cohort is good follow up on diabetes, as we do in this study; and if it was a large enough dataset and we could find such a study, we could simply hand them a diskette and say "Here, please match your controls to our Ranch Handers" and it lets you look again at diabetes. That's a possibility.

DR. STOTO: Joel, on the Vietnam Experience study, my recollection was that there were two studies, one comparing Vietnam vets to other people who served in the military, maybe the Army at the same time but not in Vietnam. And that was in fact done, results were published from that.

Then there was a second study which would have compared people who served in the Army in Vietnam in trying to establish high and low exposures and compare those to one another. But the OTA, with Mike's guidance, said "don't do that one."

DR. GOUGH: No, no, no. No, no.

We did say that, but we were ignored.

DR. STOTO: But -- that study didn't get done. That's the one that didn't get done.

DR. GOUGH: No, no, no. That was the study that was done with the 600 people, the 600 men in Vietnam, 100 or so out, and there was no evidence for dioxin exposure.

So that study was dropped because you couldn't find people -- there was no power to find people who were exposed.

Yes, the Vietnam Experience study was just simply, "Did you go to Vietnam, did you not go to Vietnam?" Those results were published, and I don't know -- I don't even know if that was discontinued. The reason for discontinuing the dioxin study is pretty clear; because we couldn't find any evidence for --.

DR. MICHALEK: We're just saying that the idea is that this would address veteran frustration.

DR. STOTO: I'm not sure which one you're talking about revising.

DR. MICHALEK: I'm not sure, either. Although we need to talk about that.

DR. GOUGH: No, but the guys who had the dioxin measurements were not participants were not participants in the Vietnam Experience study, as I recall. They did not go and have the physicals and things.

DR. STOTO: Right. That's my recollection, too. But I'm not sure about that.

DR. MICHALEK: We didn't know those details, but we can check it out.

Almost done. Yes?

DR. HARRISON: No -- I'm just saying.

DR. MICHALEK: Okay. Go.

[Photo]

Here we are, these are our buildings. Don't look very fancy, but they're very nice inside. Each one cost $300,000, and lots of high tech stuff in there. A good computer system, lots of smart people.

[Slide]

Here's one of coders. By the way, we have triple-entry quality control. Everything is coded independently and blindly by two medical coders, and then adjudicated by a third. This is unprecedented quality control in this study. Everything in this study is checked 100 percent, layers of quality control in every aspect of the study, and that's what this slide is about.

[Slide]

Here are the freezers; this is an issue. We have collected over 50,000 specimens of urine, serum, adipose tissue, and semen, and they are in the freezers. They were collected, under informed consent, through IRB approval, to address the Agent Orange issue, and we still have them in our freezers today; and that's a point of discussion for later today.

[Slide]

Our LAN, new computer equipment which makes life very efficient for us; and our new shelving for medical records, we have collected over 4 million documents on the individuals through their repeated physical examinations and their medical records that they bring to us from their family physician when they attend every physical because we ask them, plus the corresponding records on all of their children and their girlfriends and their wives that produce babies. They're all in those folders.

And all of their military health records and military records showing where they were and when during their military career.

[Slide]

And we're scanning the entire pile of paper into a system so that you can reach any document on any subject with point and click, with really great resolution. And that's what this slide is about.

DR. HARRISON: So this is all OCR?

DR. MICHALEK: No; some of it's OCR and some isn't. Many of these documents don't lend themselves to OCR because they are a doctor's scribble on a notepad. Or they were mimeographed in 1956 and they're fuzzy. But many of the reports, very clear printed reports, are OCR.

DR. HARRISON: So some of it, though, in order to actually use the data, the person getting this is going to have to sit down and transcribe it, so there's going to be an ultimate layer of errors that you don't have any control over.

DR. MICHALEK: There are many layers of information here. The physical exam --

DR. HARRISON: I'm saying your information is pristine; --

DR. MICHALEK: Yes.

DR. HARRISON: I'm just commenting--

DR. MICHALEK: Yes, with the future. What the future brings.

What you've got, if you -- maintain the only release of this study would be what's on the web page. If you have now divorced the squeaky-clean electronic data, which was used on our reports, from the patient folders, that limits the ability of anyone to do research. Because now you will see, well this kid had a defect. What was it? What did the doctor say? Well, to do that, you need to open the report.

You've got to have -- and you need to open the folder. So you have to have access to the folder, but the folder's private, because there's this privacy and confidentiality; so we have some enormous problems here with regard to preservation of confidentiality, adherence to the IRB rules about confidentiality and about the release of data and privacy.

So all of that needs to be discussed separately. Thank you very much.

DR. HARRISON: Thank you.

Any questions?

DR. GOUGH: Yes, and a couple of comments.

Joel, I found one of the slides -- I assume the slides are for technical audiences, but when you give the morbidity results, with a plus/minus, I think for a lay audience the pluses are a little misleading. Because when it says cardiovascular plus, it's --

DR. MICHALEK: I know, it's hard.

DR. GOUGH: Well, it may sound good, but the problem is, it doesn't encompass --

DR. MICHALEK: All the caveats and all the --.

DR. GOUGH: Yes.

DR. MICHALEK: The cardiovascular plus is a very complicated pic