The panel met in the Green Room, San Francisco War Memorial Building, 401 Van Ness Avenue, San Francisco, California at 8:30 a.m.,

JOYCE C. LASHOF, M.D., Committee Chair, presiding.


JOYCE C. LASHOF, M.D., Committee Chair



ANDREA KIDD TAYLOR, Dr.P.H., Committee Member

MARK BROWN, Committee Staff

HOLLY GWIN, Committee Staff


MICHAEL KOWALOK, Committee Staff








Mortality Follow-up Study of Persian Gulf



Environmental Epidemiology Service

Veterans Health Administration Department of Veterans Affairs

Principal Investigator 6


The National Health Survey of Persian Gulf

Veterans and their Family Members



Robert Wood Johnson School of

Medicine 34


Office of Management and Budget 40



Morbidity Among Gulf Veterans: A Search for Etiologic Agents and Risk Factors (studies

related to symptoms, hospitalization, and

reproductive outcomes)

Naval Health Research Center








Naval Health Research Center (continued)



Scientific Advisory Panel







1 P R O C E E D I N G S

2 8:34 a.m.


4 I think we're ready to get started. Let me just

5 make a couple of opening remarks.

6 I appreciate all of you coming today.

7 Yesterday, we spent the morning hearing from a

8 number of concerned veterans who have serious

9 problems. In the afternoon, we began our discussion

10 of epidemiologic studies.

11 We all recognize, on the Committee, that

12 epidemiologic studies are difficult ones but they

13 certainly are our first method of attack at trying

14 to uncover what may be going on among the Gulf War

15 veterans.

16 This morning, we are pleased that we are

17 going to be able to hear about the National Health

18 Survey of Persian Gulf Veterans and then also hear

19 from a series of studies being carried out by the

20 San Diego Research Group on the Seabees.

21 Before I call our first panel to the

22 podium, let me ask the members of our Committee and

23 Staff who are here to introduce themselves. I

24 should say that we are just a small subcommittee of

25 the whole Presidential Commission that is holding


1 this hearing around epidemiologic studies. We will

2 be reporting the results of these two days to the

3 full Committee at our meeting in December, assuming

4 that the debt ceiling is solved and the government

5 is functioning on December 3rd and 4th.

6 With that, let me ask Mark Brown to

7 introduce himself.

8 MR. BROWN: I'm one of those people paid

9 by the government so I'm particularly hopeful.

10 I'm Mark Brown; I'm with the Committee

11 Staff.


13 Baldeschwieler; I'm a member of the Committee and

14 I'm from the Chemistry Department at Cal Tech.

15 MS. JOELLENBECK: I'm Lois Joellenbeck

16 on the Committee Staff.


18 have said that I'm Joyce Lashof and I am Chair of

19 the Committee.

20 MS. GWIN: I'm Holly Gwin of Committee

21 Staff.

22 DR. TAYLOR: I'm Andrea Taylor; I'm an

23 Industrial Hygienist with United Auto Workers Health

24 and Safety Department in Detroit and I'm a member of

25 the Advisory Committee.


1 MR. KOWALOK: And I'm Mike Kowalok with

2 the Committee Staff.


4 Han Kang, Dr. Howard Kipen, and Ms. Allison Eydt to

5 come to the table and do our first presentation.

6 DR. KANG: Good morning. This morning I

7 am going to describe two studies that VA is involved

8 with concerning Persian Gulf veterans' health.

9 The first study I'm going to describe to

10 the Committee this morning is the Persian Gulf

11 veterans mortality follow-up study of Persian Gulf

12 veterans. This study is initiated because of the

13 concerns expressed by Persian Gulf veterans soon

14 after they come home from the theater of operations.

15 We receive many inquiries from veterans, as well as

16 Congressional Staff and Congressmen themselves, how

17 many Persian Gulf veterans die of diseases or

18 accidents.

19 There was numerous reports in the news

20 media that over 4,000 Persian Gulf veterans die of

21 Persian Gulf-related diseases of which about 1,500

22 veterans die of cancer. So we decided to conduct a

23 formal epidemiological study to compare the

24 mortality experience of those who served in the

25 Persian Gulf to the Persian Gulf era veteran.


1 (Slide change)

2 You may have already heard the

3 background information on the sequence of events in

4 the theater so I will not go into details, just to

5 let you know that, because of the different events

6 that occurred in the Persian Gulf, we may be able to

7 subgroup the study population into various exposure

8 categories. For example, those who were deployed in

9 the theater after the ground war is over, they may

10 not have been subject to anti-volatile chemical

11 agents, whereas those who were deployed at the time

12 of air war, they would have been subject to oil fire

13 fumes.

14 So, based on the history of warfare, we

15 may be able to analyze the Persian Gulf veterans by

16 their potential exposure categories.

17 (Slide change)

18 These are a list of some of the risk

19 factors that have been measured in the literature or

20 in the news media; and I will not go into details.

21 (Slide change)

22 As I mentioned earlier, the objective

23 study is to compare the post-war mortality

24 experience; I need to emphasize that. It is not the

25 mortality experience of the entire Persian Gulf


1 veterans from the beginning of the war, it is after

2 the war, to compare the post-war mortality

3 experience of all Persian Gulf War veterans to that

4 of Persian Gulf War era veterans who did not serve

5 in the Persian Gulf theater.

6 (Slide change)

7 Design is the typical historical cohort

8 studies utilizing all Persian Gulf veterans,

9 approximately 700,000, to those sample 750,000;

10 that's about one-half of the all-era veterans at

11 that time.

12 (Slide change)

13 Definition of study subjects, all

14 military personnel who served in the Persian Gulf

15 theater of operation anytime between August, '90

16 through April, 1991. The control subjects are those

17 individuals who are in the military during the same

18 time period but who did not serve in the Persian

19 Gulf. And study subjects and control subjects are

20 frequency-matched by branch and the unit status;

21 that is, active, Reserve, and National Guard status.

22 (Slide change)

23 The vital status follow-up will begin

24 when Persian Gulf veteran return alive, and control

25 subject follow-up will begin on May 1, 1991. The


1 reason we don't want to bring that follow-up period

2 to the earlier period, that the experience of

3 Persian Gulf veterans and those who stayed stateside

4 will not be substantially different. The Persian

5 Gulf veterans are isolated, they are not driving on

6 the freeway. Our understanding is that the alcohol

7 usage was almost non-existent. So, comparing the

8 Persian Gulf and non-Persian Gulf veterans prior to

9 that period will not be appropriate.

10 (Slide change)

11 The vital status ascertainment will be

12 based on VA's own computer record as well as Social

13 Security Administration death benefits record and

14 also we will utilize National Death Index for

15 control purpose.

16 (Slide change)

17 Statistical analysis, we will compare

18 the experience of Persian Gulf veterans to non-

19 Persian Gulf veterans without adjustment, it will be

20 direct comparison based on their unit components.

21 So we will be comparing the active duty personnel

22 who served in Persian Gulf against active duty who

23 did not serve in Persian Gulf, and then Reserve

24 component against Reserve component who did not

25 serve.


1 And, after -- on adjusted ratio

2 comparison with adjustment based on their, of

3 course, age, sex, the length of time in the Persian

4 Gulf area, and other demographic and military

5 variables. And then finally, we will compare the

6 mortality experience of both Persian Gulf veterans

7 and non-Persian Gulf veterans against the national

8 experience, the United States population-based, the

9 mortality statistics.

10 (Slide change)

11 These are the results as of last month.

12 We have 695,516 individuals allied with the study as

13 Persian Gulf veterans and 746,000 as non-Persian

14 Gulf veterans. After vital status ascertaining

15 using VA and also SSA files, we found 1,765

16 individuals deceased among Persian Gulf, and 1,729

17 among non-Persian Gulf veterans.

18 We are still collecting death

19 certificates but, as of last month, we were able to

20 collect death certificates for 87 percent of Persian

21 Gulf veterans and 85 percent of non-Persian Gulf

22 veterans. So we know the cause of death for 87

23 percent of Persian Gulf and 85 percent of non-

24 Persian Gulf veterans.

25 (Slide change)


1 As I say, this is based on that 85

2 percent death certificate information.

3 When we compare those who served on

4 active duty, which of over 80 percent of all Persian

5 Gulf veterans, against non-Persian Gulf veterans who

6 served on active duty, the overall mortality rate

7 are about 15 percent higher among Persian Gulf War

8 veterans. The rate ratio is 1.15, which was

9 statistically significant for Persian Gulf veterans,

10 and most of those excess mortality rates came from

11 external causes, accidents, motors vehicle

12 accidents, and other accidental deaths.

13 And the other hand, the deaths to

14 illness or diseases are much lower among Persian

15 Gulf veterans compared to non-Persian Gulf veterans.

16 The rate ratio is generally less than one. The

17 infectious diseases are 0.17; all cancer, 0.57;

18 respiratory disease, .67; and digestive disease,

19 .88.

20 And I want to draw attention to

21 infectious diseases, there are only seven

22 individuals die of infectious diseases, in contrast

23 with a lot of concerns about the effects of

24 biological warfare agents and other endemic diseases

25 in Persian Gulf area, and we only see seven cases of


1 that from infectious disease among Persian Gulf

2 veterans.

3 (Slide change)

4 Among the Reserve component, when we

5 compared Persian Gulf War veterans who came from

6 Reserve component against non-Persian Gulf veterans

7 from Reserve component who are activated and

8 deployed elsewhere, again we see similar result.

9 Their mortality due to external causes are elevated

10 although, because of the small number, that excess

11 mortality risk was not statistically significant.

12 (Slide change)

13 Unlike previous war, the Vietnam War, we

14 have approximately 50,000 women served in Persian

15 Gulf. And, when we compare them against non-Persian

16 Gulf War women veterans, their mortality experience

17 similar to the overall Persian Gulf War veterans.

18 The overall mortality risk is elevated about 50

19 percent and most of those elevated mortality risk

20 comes from external causes, accident, motor vehicle

21 accident.

22 (Slide change)

23 Finally, comparing against the U.S.

24 population, because of what we call the healthy-

25 veteran effect or healthy-worker effect, we expect


1 that their mortality experience will be much

2 favorable and here it is. Their mortality rate is

3 about half of what you expect from age, sex, race

4 adjusted national statistics. The all-cause of

5 deaths of Persian Gulf veterans SMR variable was .44

6 and, for non-Persian Gulf veterans, it's .38.

7 And it's understandable because, to be

8 in the military, you have to meet certain physical

9 criteria and also, to stay in the military, you have

10 to maintain certain physical criteria. And, when

11 they are separated, they have a better access to

12 medical care than average U.S. male or average U.S.

13 population.

14 So all the studies of veteran population

15 show favorable mortality statistics; and this is no

16 different from the previous study.

17 (Slide change)

18 For women, they have a favorable

19 mortality experience compared to the national

20 statistics but their external cause of death for

21 Persian Gulf veterans are over 1.0. Compared to

22 U.S. general female population, the Persian Gulf

23 veterans, their mortality risk for accidents and

24 motor vehicle accidents and suicides are higher.

25 (Slide change)


1 The strengths of this study that I just

2 finished describing is that we do have a veteran

3 comparison group and, also, differential estimate of

4 the vital status is unlikely because both groups of

5 veteran population went through the same vital

6 status ascertainment procedure using VA's as well as

7 SSA's files.

8 And the statistical variation will be

9 small because we are taking all available study

10 subjects, all Persian Gulf veterans, 700,000, into

11 our study.

12 Limitations of cause of death is based

13 on death certificates. Sometimes death

14 certificates, the listing of cause of death may not

15 agree with actual -- the medical record itself so

16 there is some potential for misclassification if you

17 use the cause of death information only from death

18 certificates.

19 And the most important thing is the

20 short follow-up period. Any disease that has a long

21 latent period, the follow-up period will be only two

22 and a half years, so we will not see any long-term

23 effects that originated from the Persian Gulf

24 experience and we do not have potential risk factor

25 information on both groups of veterans.


1 (Slide change)

2 In summary, since the war, Persian Gulf

3 veterans experienced a significant excess overall

4 mortality when compared to their military peers who

5 did not serve in the Operation Desert Shield/Desert

6 Storm, and that excess mortality occurred mainly due

7 to external causes, accidents and trauma, rather

8 than natural causes. And the mortality risk

9 experienced by Gulf War veterans was still less than

10 half of all expected from civilian counterpart for

11 overall causes as well as specific disease category.

12 That concludes my description of the

13 mortality study.

14 COMMITTEE CHAIR LASHOF: I think what we

15 should do now is take questions from the panel

16 concerning the mortality study, as I understand that

17 Dr. Kipen and Allison Eydt will be commenting on the

18 other study on the survey.

19 So, if the panel has any questions on

20 the mortality, let's do that now and then proceed to

21 the National Survey studies.

22 DR. TAYLOR: I want to go back to your

23 point about the death certificates.

24 How reliable is that information?

25 DR. KANG: The death certificate


1 information is very reliable, first of all, for

2 counting deaths. It is a legal document so it is

3 presumed to be a hundred percent reliable for --

4 DR. TAYLOR: But you said it was

5 different than what --

6 DR. KANG: But, for cause of death,

7 especially for cancer, there is, you know, several

8 studies comparing the cause of death information on

9 the death certificate against medical records.

10 For accidental death and external cause

11 of death, it is very reliable; but for natural

12 causes, especially for cancer, especially for

13 specific type of cancer, for example non-Hodgkin's

14 lymphoma, it is not as reliable as external cause of

15 death.

16 DR. TAYLOR: Okay, so this is just

17 looking at the death certificate.

18 Were there any attempts made to look at

19 comparisons between what appeared on the medical

20 records versus what you had in the death

21 certificate?

22 DR. KANG: Not at this point. The

23 driving force behind conducting this study is that,

24 in our experience dealing with the previous war

25 veterans, it tends to show elevated risk due to


1 external causes, the Vietnam veterans, World War II

2 veterans, Korean War veterans, the external cause of

3 death are elevated within a few years after they

4 come back from the war.

5 So the main focus -- hypotheses behind

6 this study was that we would expect to see elevated

7 mortality due to external causes. In fact, we did

8 find that from this study.

9 But, for disease with a long latent

10 period, we cannot use the death certificate to make

11 any accurate assessment of mortality risk for, for

12 example, the particular type of cancers.

13 DR. TAYLOR: Okay. So how useful would

14 this study be for delving into finding out whether

15 there was any relationship to Gulf War veterans or

16 any of the health symptoms that Gulf War veterans

17 are experiencing?

18 DR. KANG: Well, I think the conclusion,

19 at least based on preliminary data, is that their

20 mortality risk due to natural causes, that is due to

21 diseases, are lower than their counterpart who did

22 not serve in the military.

23 We have a lot of concern about the

24 illness and diseases that related to Persian Gulf

25 veterans but at least it hasn't shown yet from their


1 mortality experience.

2 DR. TAYLOR: And I don't know if you

3 would expect that it would be at this point, though.

4 DR. KANG: It will be too short.

5 DR. TAYLOR: Too short; okay.


7 to do -- to follow this group and --

8 DR. KANG: Yes.


10 periodically recheck? I don't know how often one

11 ought to but, several years downstream, taking

12 another look seems to be indicated.

13 DR. KANG: Since we already have a core

14 developed, the periodic follow-up updating the

15 mortality experience downstream will not be much

16 problem so we plan to do that in, say, every two or

17 three years.


19 question I have is, as you point out, one of the

20 limitations was that you don't have specific risk

21 factors. Now, granted, since the overall mortality,

22 except for the excess external causes are not

23 different, still I wonder whether, once we get the

24 troop location information, which we expect to get

25 from the Department of Defense within the next month


1 or two, whether it will be worth going back and

2 looking at some of the specific locations where

3 we've had anecdotal presentations to this Committee

4 of high number of deaths in specific units, whether

5 we ought to take a look at those.

6 DR. KANG: We would like to do that.

7 When that information is available, we can quantify

8 the Persian Gulf veteran into the time and space in

9 the theater and see whether there is any difference

10 in the mortality experience based on their troop

11 movement and the time of performance.


13 look forward to that, too.

14 Are there any other questions?

15 Lois?

16 MS. JOELLENBECK: A comment and a

17 question.

18 The comment was, it looks from the data

19 as though, between deployed and non-deployed, there

20 really is a healthy-worker effect, a healthy-soldier

21 effect, which brings us back to something we touched

22 on lightly yesterday which was, when we're using the

23 deployed-but-not-to-the-Persian Gulf as a comparison

24 group, how good a comparison group is it.

25 DR. KANG: Okay.


1 MS. JOELLENBECK: And this just might

2 make us aware that, in fact, the deployed-to-the-

3 Persian Gulf might be expected to be healthier.

4 DR. KANG: I don't know what kind of

5 screening there was to select individuals to be sent

6 to the Persian Gulf; I guess the DOD can answer

7 that.

8 But this particular comparison that I

9 presented here, if you look at the control group,

10 you know, we can say non-Persian Gulf veterans. For

11 those who served in Reserve units, there are two

12 parts of control that we are utilizing for this

13 study as well as the survey that I'm going to

14 describe later, that there are Reservists who are

15 activated and deployed somewhere, either stateside

16 or overseas, but not Persian Gulf area. And there

17 are Reservists who are in Reserve but not called up.

18 So I like to compare the effects of the

19 activation, or deployment itself, by comparing the

20 Reservists who are activated and deployed versus the

21 Reservists who are not activated at all. And this

22 particular comparison that I presented is the

23 comparison between Reservists who are activated and

24 sent to Persian Gulf, and Reservists who are

25 activated and sent elsewhere.


1 And I don't know whether there is a

2 particular physical -- to send one person to Persian

3 Gulf and one person to Germany; I don't know that.

4 But at least that kind of comparison will minimize

5 that healthy-soldier effect.

6 MS. JOELLENBECK: My question was, at

7 the beginning of your talk, about how you could

8 analyze, maybe by different time periods, that

9 people were in the Gulf. Have you moved ahead to do

10 that?

11 DR. KANG: We'll do that. We just

12 finished the preliminary analysis; we haven't done

13 any subgroup analysis based on the time period when

14 they are in the Gulf area or their military

15 occupation. And we'd like to compare those who have

16 combat MOSC, the Military Occupational Specialty

17 versus the support troops, compared to based on the

18 branch because their experience may not be the same,

19 either environmental or psychological experience.

20 We would like to do all that analysis in the future.


22 DR. BALDESCHWIELER: In your comparisons

23 with the civilian non-military operation, are those

24 cohorts age-matched?

25 DR. KANG: Yes.


1 DR. BALDESCHWIELER: Because it seems to

2 me age is a critical factor.

3 DR. KANG: Oh, sure.

4 DR. BALDESCHWIELER: And are there

5 systematic differences in age when you compare the

6 Reserves versus the active duty troops that were --

7 DR. KANG: The Reservists are older; I

8 think they are two or three years older than active

9 duty troops.

10 DR. BALDESCHWIELER: And so, when you

11 compare the era veterans versus the Gulf War

12 veterans, is there a systematic age difference

13 between those total groups?

14 DR. KANG: Well, we try to account for

15 that by comparing Reservists against Reservists, the

16 age are about the same, almost identical. And, when

17 we do SMR, we are adjusting for age, sex, race, and

18 calendar year of death because some cause of deaths

19 are going up, for example, say lung cancer on women

20 in general are going up without knowing what causes

21 that.

22 DR. BALDESCHWIELER: I can imagine that

23 there is a systematic selection of age among those

24 -- for example those Reservists who are called up to

25 serve and those who are not. I would expect the


1 younger people had a higher rate of call-up.

2 DR. KANG: That's why I'm just comparing

3 those who are called up. Both groups are called up,

4 one sent to Persian Gulf and one sent to somewhere

5 else.


7 DR. KANG: And, when I look at the

8 demographic characteristics, they are identical in

9 terms of age and race.

10 DR. BALDESCHWIELER: One extraordinary

11 feature of your results is the difference in

12 infectious disease, the numbers are very small for

13 deaths in those particular categories. But I wonder

14 if that reflects the selection on the basis of

15 positive versus negative HIV?

16 DR. KANG: In the veteran population in

17 general or --


19 recollection that they were screened for HIV.

20 DR. KANG: Right. I think the Armed

21 Forces screen for HIV at the entrance and as well as

22 periodically while they're in the service. So they

23 are --

24 DR. BALDESCHWIELER: Wouldn't there be

25 HIV-positive soldiers in the era but-not-sent-over


1 group?

2 DR. KANG: I don't have that

3 information; I can't find that out.


5 differential or is HIV-positive a cause of rejection

6 for military service or Reserve service altogether?

7 I simply don't know.

8 DR. KANG: I don't know either.


10 answer to that, go ahead.

11 DR. GREGORY GRAY: It's very true we

12 have people who are infected with HIV who remain in

13 the service and there is a judgement call regarding

14 the progression of their disease. There is a period

15 where they actually leave the service and if so --

16 and depending on their status they are kept

17 stateside where they can obtain excellent health

18 care.

19 So you're right in the sense that the

20 differential representation with respect to HIV

21 serology. But I want to remind you that the number

22 of people infected with HIV in the DOD is very very

23 small compared to the number of people in the

24 general population.



1 Anything else or can we move on.

2 MR. BROWN: May I ask one?


4 MR. BROWN: Did I understand you, Dr.

5 Kang, to say that there is a precedent for observing

6 this effect of greater accident -- greater mortality

7 rates due to accidents in returning veterans from

8 previous wars? And are the effects that you saw

9 comparable to the -- in terms of their magnitude to

10 these other situations?

11 DR. KANG: I don't know where there is a

12 coincidence or not. CDC conducted the mortality

13 study of Vietnam veterans to compare Vietnam ground

14 troops against the non-Vietnam ground troops which

15 was published in 1986 and they show .7 percent

16 elevated overall mortality among Persian Gulf

17 veterans and their risk ratio is lower for natural

18 cause and then, of course, the risk ratio is higher

19 for external cause including --

20 MR. BROWN: It's similar.

21 DR. KANG: So it is not -- it is

22 consistent with the previous war veterans' mortality

23 study.


25 whether any of those external causes or accidents


1 were related to alcohol abuse and drug problems and

2 depression and psychological problems that followed

3 the war?

4 DR. KANG: Well, we didn't have access

5 to the medical records so we can't say anything

6 about that yet. But the CDC mortality study that I

7 just finished describing, they did go back and

8 collect the medical records of those accidental

9 deaths except for motor vehicle accidents. I think

10 they were able to add two or three additional cases

11 to suicide based on the record received.

12 So I don't think there would be many

13 hidden suicides among accidental deaths.


15 I think we need to move on to ask you

16 now to discuss the National Survey of Persian Gulf

17 War veterans, which is just getting under way.

18 DR. KANG: Yes.


20 don't have results, you are just going to tell us

21 about how it's playing and what we can hope to get

22 from it.

23 DR. KANG: Almost a year and a half ago,

24 there was a workshop at NIH in April. The panel

25 members of that group committee were asked to review


1 the relevant information and make a recommendation

2 what future research is needed.

3 (Slide change)

4 The NIH panel recommended that more

5 accurate estimate of the symptom prevalence be

6 established. They had been hearing a lot about, you

7 know, high proportion of returning persons

8 complaining of various symptoms. Our experience

9 dealing with Persian Gulf Registry, House Registry,

10 it is a self-elected group coming to the aid,

11 receiving medical examinations.

12 The proportion of individual with

13 fatigue is in the range of 20 percent, with the skin

14 problem, 20-some percent; and this all self-selected

15 group, there is no good estimate of symptom

16 prevalence as well as the medical conditions. Both

17 NIH panel recommend that we should do that, we

18 should either conduct a survey of entire 700,000

19 veterans or do representative samples and then

20 conduct a survey.

21 So that was NIH panel recommendations.

22 And the VA accepted recommendation and decide to do

23 National Survey.

24 And then the following November,

25 Congress passed a law, Public Law 103-446, actually


1 direct the Secretary of Veterans Affairs to conduct

2 the House survey. So those two instance are driving

3 force behind what we designed for the National

4 Survey.

5 (Slide change)

6 We have three objectives of purpose, to

7 estimate and compare prevalence of symptoms and

8 health outcome among Persian Gulf veterans and non-

9 Persian Gulf veterans; and also to estimate and

10 compare prevalence of their reproductive health

11 among spouses and birth defects among children of

12 Persian Gulf veterans and non-Persian Gulf veterans;

13 and then finally, to the extent of -- and I need to

14 emphasize "to the extent possible" because most of

15 the exposure information we will collect will be

16 self-reported and non-verified.

17 So, to the extent possible, prevalent

18 relations between selected symptoms and health

19 outcome and then certain environment exposure in the

20 Gulf area.

21 (Slide change)

22 Study subjects, we sampled 15,000

23 Persian Gulf veterans from the cohort that we

24 developed and used for the mortality study and an

25 equal number of non-Persian Gulf veterans from the


1 population that we select for the mortality study.

2 (Slide change)

3 Study design, we will use questionnaire;

4 it will be mail questionnaire supplemented with a

5 telephone interview for the non-respondents and then

6 finally, physical examination on a sample of

7 veterans and their family members.

8 And also we will supplemental validate

9 the self-reported exposure and health outcome of

10 data against available military NDA records.

11 (Slide change)

12 For the mail survey, we will send out

13 the initial notification letter, followed up with
14 the questionnaire and then postcard reminder and

15 then replacement questionnaire and then another

16 replacement questionnaire.

17 Because of the expected low response

18 rate based on mail survey, we will do all we can to

19 increase response by sending out the reminders and

20 follow-up questionnaires and then also supplemental

21 telephone interview for a sample of non-respondents

22 to ascertain non-respondents bias and then physical

23 examinations.

24 (Slide change)

25 The time table, the Phase I postal


1 survey will take us about a year, 12 months.

2 (Slide change)

3 The telephone interview will take some

4 other 18 months. During that time, we will validate

5 the medical records of random sample 1,000 from

6 Persian Gulf and 1,000 from non-Persian Gulf groups,

7 compare their medical records, both from civilian

8 facility as well as DOD and VA facility.

9 And then final physical examination from

10 the same 1,000 respondents from Persian Gulf group

11 and that non-Persian Gulf group, veterans groups,

12 will be conducted.

13 This table show this division of study

14 subjects and control by Persian Gulf service status,

15 as well as the unit component status. Our expert

16 panel recommend that we do over-sample -- we should

17 over-sample Reservists and the National Guard

18 servicemen and also women, for those three groups

19 are oversampled. Altogether, 15,000 Persian Gulf

20 veterans, 3,000 of those 15,000 will be women.

21 (Slide change)

22 The strength of this proposed National

23 Survey, based on postal questionnaire is -- it is

24 not a very expensive propagation compared to either

25 telephone survey of all 30,000 or in-person survey


1 of all 30,000. And because this postal survey will

2 be much shorter information collection time compared

3 to other methods, and we can have a very large

4 number of individuals for the study. And also wider

5 geographic area.

6 And respondent will have the time to

7 review and compare their -- pull out their medical

8 records or any records they need to answer the

9 questions.

10 So those are the strengths of the postal

11 questionnaire method.

12 The limitations are very obvious. We

13 expect lower response rate, 40 to 60 percent

14 compared, say a telephone survey which, if it is a

15 good telephone survey, range around 80, 90 percent.

16 The questionnaire has to be very short

17 because, you know, we don't expect many people to

18 spend hours and hours to filling out the

19 questionnaire, otherwise it will go down the

20 wastebasket. So there is a limited number of

21 questions that we can ask. And the question has to

22 be very simple and straightforward. And it doesn't

23 give a chance to ask follow-up questions.

24 So those are the limitations of the

25 methodology that we are proposing here.


1 (Slide change)

2 This is the table describing sample size

3 we require for certain statistical power given

4 background prevalence rate. For example, what you

5 are interested in is the symptom that has five

6 percent prevalence rate among your comparison

7 population, for example, say, slip disturbance.

8 To have a 90 percent statistical power

9 when, in fact, the difference is say over 50

10 percent, you will need 1,966 individuals in each

11 study. And we have 30,000 people in each study.

12 So we have adequate statistical power

13 for the prevalence, the symptom prevalence up to

14 five percent. But if your symptom prevalence is two

15 percent, for same statistical power with the same

16 relative risk, you will need 5,116 people for 90

17 percent statistical power.

18 So statistical power, for our study, is

19 fairly good because of the large sample size we are

20 proposing.

21 (Slide change)

22 And, of course, if you have a higher

23 prevalence, like if 15 percent, which is what we

24 observe among the Persian Gulf veterans who come for

25 Registry examination for fatigue you will need a


1 very small number, only 567 people in the study.

2 (Slide change)

3 This is what transpired up to this

4 point. As I described earlier, after April 27 NIH

5 Technology Assessment Workshop, we proposed a study

6 to Persian Gulf Expert Scientific Committee in July

7 meeting. They accept that and asked us to develop a

8 full study protocol which we presented in November

9 meeting.

10 And, because of the need for having

11 standing oversight committee and because of the

12 difficulty of establishing advisory committee

13 because there is a moratorium on federal advisory

14 committee, the advisory committee decided to have a

15 subcommittee overseeing the proposed National

16 Survey. So they established the oversight committee

17 as subcommittee of the parent VA Persian Gulf Expert

18 Scientific Committee.

19 So we had our first meeting in November

20 and, in January, the Secretary gave us approval to

21 go ahead with the Study. In March, we had another

22 meeting and then, in April, we received final

23 approval for protocol as well as the questionnaire

24 that we proposed.

25 In May, we applied for OMB approval for


1 the questionnaire and, on September 20, we received

2 the approval. On October 11th, we sent out pre-

3 notification letters and, today, the actual

4 questionnaire is being mailed out.

5 That's where we stand in terms of

6 progress.

7 (Slide change)

8 The individuals who serve on the

9 oversight committee, Dr. Stolley is the Chair, and

10 Dr. Bascom, Dr. Leon Gordis. Dr. Gordis is not a

11 member of the parent Expert Scientific panel but he

12 is a member of the subcommittee because of his

13 expertise in epidemiology. And Dr. Howard Kipen,

14 who is also a member of the full committee, serving

15 on the oversight committee, and Dr. Jim Melius.

16 Those five individuals are serving as oversight

17 committee members.

18 That concludes my --


20 much and we are very pleased that Dr. Kipen is here

21 to talk from the oversight committee and he will

22 comment at this point.

23 DR. KIPEN: I'd like to thank the

24 Committee very much for being here and I will

25 apologize in advance for the art work on my


1 overheads which, because of the timeline done on the

2 airplane, which was a smooth flight, fortunately.

3 (Slide change)

4 As Dr. Kang said, I am a member of the

5 oversight committee although, again, because of the

6 timeline, I have reviewed our minutes, I have not

7 consulted with any of the other committee members

8 prior to this presentation.

9 The aspects of the study that we

10 concentrated on in our two face-to-face meetings and

11 one conference call were a review of the goals of

12 the project and I will, for each of these things, I

13 will tell you what in particular we recommended in a

14 moment, and then also various aspects of the methods

15 which included the design and protocol, various

16 aspects of statistical analysis, considerations of

17 power, study recruitment -- meaning the invitation

18 letter -- the format of various questions, and then

19 finally the content of some questions.

20 If I could have the next overhead,

21 please.

22 (Slide change)

23 What I have listed for you on this slide

24 are the areas in which I thought the study was

25 significantly modified by our input of where the VA


1 investigators were significantly responsive to

2 things that we talked about.

3 First of all was the idea that, in terms

4 of the goals, that the goal of doing etiologic

5 analyses in determinations be de-emphasized because

6 of the limited ability to capture information on

7 exposures of interest, or primarily because of that.

8 In fact, it has been de-emphasized, as Dr. Kang

9 said.

10 We asked, in particular at the first

11 meeting, for some clarification and increased detail

12 on the statistical methods that he proposed to use

13 and that was provided and the committee was

14 satisfied with the robustness of those methods.

15 We were concerned that initially the

16 sample size for the Phase II follow-up telephone

17 interviews of 500 persons, that it would not provide

18 adequate power for assessing prevalence of

19 relatively rare or less than even five percent

20 outcomes in there. And, because of that, we

21 suggested to Dr. Kang that an attempt be made to

22 increase that number to 4,000.

23 He was reluctant at first, I think

24 because of concerns of the cost of the telephone

25 interviews. However, I believe that it has now been


1 increased to 4,000 and so we're glad that those

2 resources were recovered.

3 We suggested some revisions of the

4 recruitment letter to clarify for the recipients or

5 the potential subjects that confidentiality would be

6 protected and also to try to appeal to what we

7 thought would be some of their nobler instincts

8 rather than compliance with Public Law 103-446. And

9 we think the letter has made some improvement in

10 doing that.

11 We suggested some changes in order of

12 questions that we thought would be more well-

13 received by subjects, some changes in how -- one of

14 the characteristics of the questionnaire that I

15 particularly like is the way the -- for symptoms, we

16 ask when they began and we ask subjects to clarify

17 whether it was before the Gulf, during service in

18 the theater, or afterwards. And how that's

19 formatted was a little bit complex and I think it

20 kind of came out quite nicely.

21 We made a number of detailed suggestions

22 on the reproductive section having to do with

23 identification of partner or spouse for a given

24 child and just how this would be recorded in a

25 useful manner. We suggested some revisions in -- in


1 fact, a different version of the review of systems

2 questionnaire and that was incorporated.

3 We suggested some clarifications of the

4 definition of the term "severe" for quantifying or

5 assessing how severe symptoms were and those

6 suggestions were also incorporated, although I

7 understand, during subsequent negotiations with OMB,

8 some of that has been changed, or at least the

9 moderate category of symptom severity has been

10 eliminated.

11 We suggested, in particular, some

12 questions to be added to assess the prevalence of

13 chronic fatigue or chronic fatiguing illness among

14 recipients of the questionnaire. And many of those

15 were added. We suggested the deletion of some

16 questions on chemical sensitivity which had been

17 proposed in initial versions of the questionnaire

18 but which we did not think were among the most

19 robust questions available for assessing that

20 symptomatic phenomenon and did suggest a replacement

21 question about chemical sensitivity be substituted;

22 and that is on the questionnaire.

23 Could I have the next slide, please?

24 (Slide change)

25 In terms of the strengths, Dr. Kang went


1 over most of those. I, and I think the rest of the

2 committee, feel that this is the best thing that I'm

3 aware of that's going out there for generating

4 prevalence of -- estimates of prevalence of symptoms

5 in the deployed cohort relative to appropriate

6 control groups, the ability to divide the comparison

7 groups into those who were deployed to the Gulf and

8 those who were merely activated, we see as a big

9 strength. The record review validation, in terms of

10 looking at hospitalizations and perhaps some other

11 end points will be quite useful.

12 And finally, the estimates now of non-

13 respondent bias with their increased size should

14 also be a big strength of this compared to many

15 other efforts of a similar type.

16 The limitations are that we still don't

17 have all the power we would like for some of the

18 things that concern a lot of people, certainly

19 individual reproductive outcomes and things like

20 that. But I think that, for looking at the

21 symptomatic illnesses, which this is best designed

22 to take a look at, I think we're going to be doing

23 okay in terms of power.

24 Exposures continue to be self-reported

25 and I'm sure your Committee is very familiar with


1 the problems with that. We will, of course, have

2 problems with lapse of memory as well as recall bias

3 among symptomatic people or people who have

4 experienced adverse outcomes.

5 And then finally, limitation will be

6 that we will not really be able to estimate the

7 prevalence of symptoms characteristic of actual full

8 chronic fatigue syndrome or of chemical

9 sensitivities because we did not use as many

10 questions as some members of the committee thought

11 would be useful for pursuing those lines of

12 investigation.

13 That concludes my prepared remarks.


15 much, Dr. Kipen.

16 I think we will move on and ask Ms.

17 Allison Eydt from the Office of Management and

18 Budget to comment and then we will open it up for

19 questions from the panel.

20 MS. EYDT: Good morning members of the

21 Presidential Advisory Committee and researchers.

22 It is a pleasure and an honor to discuss

23 today the Office of Management and Budget's role in

24 reviewing federally-sponsored surveys on Gulf War

25 veterans' illnesses.


1 In particular, your staff has requested

2 that I highlight OMB's review of the Department of

3 Veterans Affairs survey entitled "National Health

4 Survey of Persian Gulf Veterans and their Family

5 Members."

6 Before discussing Dr. Kang's survey, let

7 me begin by providing some brief background on OMB's

8 role in reviewing federally-sponsored surveys.

9 I am a Senior Desk Officer in OMB's

10 Office of Information and Regulatory Affairs and

11 have worked in health regulation and research for

12 the last ten years at OMB on programs that include

13 the VA, Department of Defense, CHAMPUS, Medicare,

14 and Public Health Service accounts such as the

15 Agency for Health Care Policy and Research and the

16 National Center for Health Certificates.

17 As an OMB Desk Officer, I carry out

18 OMB's responsibilities under the Paperwork Reduction

19 Act in concert with other OMB Desk Officers,

20 statisticians, economists, budget analysts, and

21 other analysts from sister agencies such as the

22 Office for Science and Technology Policy. I serve

23 as the public point person for review of agency

24 efforts to ensure compliance with the Act.

25 Certain provisions of the Act and its


1 regulations have special import for the surveys we

2 have discussed in this public hearing. For example,

3 fundamental objectives of the Act are: to

4 coordinate; integrate; and to the extent practicable

5 and appropriate, make uniform federal information

6 resources management policy and practices as a means

7 to improve the productivity, efficiency, and

8 effectiveness of government programs; and to improve

9 the quality and use of federal information to

10 strengthen decision-making, accountability, and

11 openness in government and society.

12 Further, the statute requires that the

13 Director of OMB maximize the practical utility of

14 and public benefit from information collected by or

15 for the federal government.

16 The PRA regulations define practical

17 utility as the actual, not merely the theoretical or

18 potentially usefulness of information to an agency,

19 taking into account its accuracy, validity,

20 adequacy, and reliability, and the agency's ability

21 to receive and process the information it collects.

22 The VA was submitted to OMB earlier this

23 summer. Initially, OMB reviewed the VA study in

24 isolation of other Gulf War studies and on its own

25 merit. OMB staff considered the primary objectives


1 of VA's research and whether they were consistent

2 with VA's unique statutory and regulatory missions,

3 the proposed sample size and methodology, VA's

4 expected response rate, and plans for data

5 validation and respondent follow-up, et cetera.

6 In its supporting statement, the VA was

7 frank as to the research challenges in this area and

8 outlined potential pitfalls. OMB appreciated VA's

9 foresight and directness in exploring these issues

10 and ultimately approved this survey with the

11 expectation that many of these limitations would be

12 caveat on any dissemination of research findings.

13 The issues that concerned OMB most were

14 related to research coordination. As a result of

15 its Paperwork Act mandate and oversight role of

16 numerous executive agencies, OMB staff became aware

17 that the VA National Study was one in a series of

18 agency efforts. To meet conditions for OMB approval

19 under the law, it was imperative that the VA study

20 be coordinated sufficiently with other studies. In

21 particular, during the course of our review, the

22 Centers for Disease Control Iowa Study was submitted

23 and we learned of a promising VA study that would

24 pursue exposures in more detail.

25 Because OMB staff knew of these studies


1 in the course of our formal review, under the PRA,

2 the Paperwork Reduction Act, we coordinated the VA

3 Study with these efforts and vice versa. Within our

4 time constraints, we did the best we could to move

5 the studies in this direction.

6 Because of its reasonable design, scope,

7 and national representativeness, we focused on the

8 VA National Study as a valuable benchmark study for

9 future prevalence studies. Under this premise, we

10 compared its core question modules identifying

11 symptoms, medical conditions, and exposures with the

12 other two studies and targeted areas of deviation.

13 We felt it was important to use this VA

14 review as an opportunity to develop a standard set

15 of symptoms and, to a lesser extent, medical

16 conditions and exposures that could be validated by

17 other agency regional and localized studies as well

18 as defining its research progressive.

19 When OMB shared copies of these draft

20 studies between CDC, HHS, and DOD, we were pleased

21 to learn that the agency researchers were familiar

22 with each other's work. The agencies justified any

23 differences in terms of commitment to more in-depth

24 pursuits of certain issues, preferences for

25 different survey modes or methodologies, and


1 different agency cultural perspectives.

2 OMB continues to recognize the

3 rationales underlying survey differences and the

4 importance of intellectual freedom and exploration.

5 However, OMB was concerned that, if certain

6 questions were not consistently asked in all major

7 surveys, it may be impossible to control for

8 differences in research findings that would arise

9 from intricate differences in instrument design.

10 To strike this delicate balance between

11 using studies for validation versus exploration of

12 new realms, OMB decided to work with the interested

13 agencies to create core sets of symptoms, and to a

14 lesser extent, medical conditions and exposures, by

15 proving these differences between the VA survey and

16 the CDC survey. The resulting core symptoms list

17 will be incorporated into the VA prevalence survey,

18 the Iowa Study, and hopefully future studies that

19 begin exploring exposure and causation with a

20 symptom screen.

21 Of course, future studies would be free

22 to explore other issues appropriate to their efforts

23 consistent with meeting the Paperwork Reduction Act

24 standards for approval.

25 In examining the differences between


1 surveys, OMB negotiated several amendments that OMB

2 hopes will facilitate its use as a benchmark

3 prevalence survey. For example, rather than focus

4 on symptoms identified in the last week, the VA

5 study now determines whether symptoms have been

6 experienced in the last month consistent with CDC

7 recommendations.

8 VA added or revised existing symptoms on

9 their list to be consistent with CDC's inventory.

10 VA added and amended several exposure categories

11 that have been of interest to veterans, including

12 exposure to dead animals, chemical gear protection,

13 and SCUD missile explosions within a mile.

14 It is important to note that, on a

15 separate trek, OMB worked with CDC to make similar

16 conforming changes to the Iowa Study.

17 So what does the VA review mean for

18 future survey efforts that must receive OMB

19 clearance before they can be fielded? First, to the

20 extent that agencies and other formal coordinating

21 bodies aggressively can address the issues of where

22 standardization and coordination are critical to

23 achieving refinement in research and valid results,

24 all these review process will be simpler.

25 OMB does not approve the coordinating


1 board's overall research strategy. But, to the

2 extent that this strategy does not reflect

3 coordination, it satisfies the growth of the

4 Paperwork Reduction Act. OMB must ensure that this

5 is done to the extent possible because OMB review

6 typically is at the end of the agency survey design,

7 procurement, and contracting and budgeting streams.

8 It would be sensible for maximum

9 coordination to occur earlier before agencies are

10 locked into their own instruments, methodologies,

11 and schedules. If this coordination is not apparent

12 to OMB in future reviews, however, it must continue

13 to perform this function as required by law.

14 An immediate example for consideration,

15 many of the studies that are intended to fulfill the

16 first and most basic research question pertaining to

17 symptoms prevalence. Studies that focus on

18 prevalence or have a foundation in measuring

19 prevalence should justify by deviation from the

20 symptoms categories in the VA study.

21 OMB is not saying that additional

22 categories or some minor changes will not be

23 allowed. But the agency must be able to justify how

24 responses to each category can be related to NEVA

25 national findings in a valid and reliable manner.


1 Agencies that desire to deviate from basic VA core

2 symptoms list items must explain how these

3 deviations will have utility and contribute to firm

4 research findings on the incidence and causation of

5 Gulf War illnesses.

6 To put it simply, OMB's concern that too

7 much research could lead to inconsistent research

8 findings that conveniently cannot be resolved

9 because we are left with comparing apples and

10 oranges. OMB discourages unnecessary redundancy but

11 encourages consistency that leads to valid and

12 reliable findings and the refinement of research

13 direction.

14 Several pragmatic lessons that the

15 agencies can learn from our VA review: First,

16 future studies that must receive OMB clearance

17 should adopt, to the maximum extent feasible, the VA

18 symptoms list and to allow, to lesser extent, the

19 medical conditions list and exposure list. Agencies

20 must be prepared to identify any such deviations up

21 front and explain the research justification for

22 such deviation.

23 For example, an agency may explain that

24 its more-detailed categories could be collapsed into

25 one of the VA symptoms categories. In particular,


1 regional or local studies relying on symptoms screen

2 and prevalence measurement should be designed to

3 ensure that comparisons can be made to the VA

4 study's national results.

5 CDC worked with OMB to accomplish this

6 with its Iowa Study. Future studies may want to

7 build off the VA study as a screener instrument and

8 even use it as a sample base.

9 Second, the OMB-supported statements

10 should explain in detail how the agency has

11 coordinated its studies with other efforts designed

12 to answer the same research, question for question.

13 This may prevent OMB from reinventing the wheel and

14 will help us provide a more expeditious review.

15 Third, if the study is one phase of

16 several research protocols, agencies should provide

17 OMB with as much detail as possible regarding the

18 substance and methodology of future phases. OMB

19 cannot evaluate the practical utility of the

20 screener instrument unless it understands the entire

21 project and how its components together will address

22 research objectives.

23 Agencies should know if this study is

24 subject to OMB review and plan for the review period

25 in their schedule. All federally-sponsored studies


1 asking standard questions of ten or more respondents

2 are subject to OMB clearance, the joint-agency OMB

3 review can be expected to take from 90 to 120 days,

4 and finally, if the agency has completed its design

5 and supporting materials, OMB recommends that it

6 shares these materials in draft with us while the

7 public 60-day comment period is in progress. Early

8 discourse on issues may facilitate early resolution

9 and more expeditious OMB review.

10 MR. LEIGHTON: Can you --

11 MS. EYDT: Conclusion --


13 MS. EYDT: OMB must satisfy a broad

14 mandate for ensuring maximum usefulness and

15 coordination in federal research. In reviewing

16 individual surveys, OMB attempts to appreciate and

17 evaluate prior efforts to fit discreet surveys into

18 a meaningful, comprehensive federal research

19 strategy. To the extent that this work has been

20 accomplished, it makes OMB's job simpler.

21 But, if this coordination does not

22 appear to have been accomplished with appropriate

23 attention to detail, OMB may work with agencies to

24 amend individual surveys to facilitate such

25 coordination. OMB strove to meet its coordination


1 responsibilities in this manner in the context of

2 its review of the VA National Prevalence Survey.

3 OMB looks forward to working with all

4 interested parties in ensuring that future Gulf War

5 surveys produce valid and reliable data that serves

6 the public interest.

7 Thank you and I welcome your questions.


9 much. We will now open it up for questions to any

10 one of the members of the panel.

11 Andrea?

12 DR. TAYLOR: Thank you.

13 I'm pleased to see that OMB is actually

14 helping to coordinate efforts between, I guess, the

15 Iowa Study and the VA Study. And I'm familiar with

16 OMB from working with OSHA and how long it takes for

17 a standard to be passed and I know this was a little

18 different.

19 However, I guess I'll go back to the

20 question of, in the process of reviewing research

21 that is going on and the proposals for research that

22 you are going to be doing as far as questionnaires

23 and other, the process you mention will take 90 to

24 120 days.

25 Now I've heard that before and I've


1 heard it extended far beyond that. So I'm just

2 trying to get a sense of what that actually means,

3 more long-term.

4 MS. EYDT: Right. Hopefully, I won't

5 sound too bureaucratic, so I'll try to simplify it a

6 little here.

7 But the process begins with a notice

8 that's published by the agency in the Federal

9 Register. That notice commences a 60-day review

10 that's managed by the agency, comments are sent to

11 the agency. After those 60 days, the agency reviews

12 the comments and decides whether amendment would be

13 preferable or necessary.

14 At that point, after they make those

15 changes, they put in another notice in the Federal

16 Register saying "we're sending it to OMB now" and,

17 at that point, we would commence our review. We

18 have to hold the instrument for a minimum of 30

19 days, so we have a minimum 30-day review of any

20 instruments. And that can be extended for up to 60

21 days.

22 So that's where I get the 90 to 120

23 days; that's actually -- it's quite conservative

24 assuming -- that assumes there will be no changes

25 made by the agency after the 60-day comment period.


1 DR. TAYLOR: Going back to my previous

2 experience with OMB, I know that it takes usually

3 longer than that.

4 MS. EYDT: I can't speak to your

5 previous experience.

6 DR. TAYLOR: All right.

7 MS. EYDT: But we are compelled, under

8 our regulations, at 5 CFR 1320, to take no longer

9 than a certain period of time, the time frame that

10 I've laid out.

11 I think some of the frustration may

12 result, also, from the review period within agencies

13 before they even get to the first step and so it's

14 all -- because it -- they're doing that because of

15 us, it all gets added in together.

16 But we are compelled under our

17 regulations to take no more than the 120 days in the

18 whole process.

19 DR. TAYLOR: Okay. I have one other

20 question of Dr. Kipen and Dr. Kang, with the

21 research.

22 You mentioned that questions related to

23 chronic fatigue syndrome and multiple chemical

24 sensitivity were either changed or reduced -- the

25 number of questions? I'm not certain.


1 Can you go back over what happened in

2 that case with chronic fatigue syndrome and the

3 questions related to multiple chemical sensitivity?

4 DR. KANG: Because the study is the

5 symptom prevalence survey, we didn't want to focus

6 on any particular syndrome and I alluded to earlier

7 that one of the limitations of the postal survey is

8 that the length of the -- the questionnaire has to

9 relatively short.

10 So we were struggling with, you know,

11 whether we should lengthen to capture more

12 information or shorten it and have a higher response

13 rate. It's a trade off between the two and there

14 were extensive discussion among the oversight

15 committee and I think the final recommendation is

16 that they will leave it up to the principal

17 investigator to make that judgment.

18 So we compromised and took, not the

19 entire questions dealing with the multi-chemical

20 sensitivity and chronic fatigue syndrome, but some

21 of the questions dealing with chronic fatigue and

22 multi-chemical sensitivities so that, in the future

23 -- in fact, there is a marked difference between two

24 groups and somebody can explore that with another

25 study.


1 But we thought that this study cannot be

2 a study for all health outcomes or -- that's where

3 we stand.

4 DR. TAYLOR: Okay.


6 that, obviously, the response rate in mail

7 questionnaires is a problem and you have increased

8 the number of phone interviews you will do to try to

9 deal with that.

10 But you also indicated that, from the

11 physical examination, you would be doing a -- and

12 I've forgotten the number that would be done -- but

13 those were all follow-ups on those who responded and

14 it would be a certain number of deployed and non-

15 deployed.

16 DR. KANG: Right.


18 there is any chance of getting those who don't

19 respond to the questionnaire that you might have

20 been able to reach by phone to convince them to come

21 in for a physical examination so that you have some

22 better control on the non-respondents and whether

23 you are getting too much of a bias group from those

24 who responded compared to the non-?

25 DR. KANG: Perhaps we didn't really


1 explore that possibility but we thought that, if

2 individuals choose not to even respond to the simple

3 postal survey, it would be extremely difficult to

4 have them come in for physical examination.

5 So that was a pure assumption.


7 DR. KANG: So our current proposal to

8 not include physical examination of all those

9 individuals who refused to participate.


11 suspect, when you get to look at your non-

12 respondents and how -- whether they differ

13 significantly from the others, it may be worth

14 thinking about whether you could convince people who

15 don't want to fill out a questionnaire but, if given

16 an opportunity to have a complete physical exam at

17 some place close to them and so forth, might well

18 avail themselves of it. And it might be another way

19 to get at it. I'm not sure, but --

20 DR. KANG: We don't have any reason not

21 to include, it was a practical decision that we

22 made.


24 whether Dr. Kipen has any thoughts on that either,

25 just one I asked --


1 DR. KIPEN: I guess I would have to echo

2 Dr. Kang's concern about the likelihood that they

3 would accept it. And then you have to wonder a

4 little bit about what you're going to get out of it

5 in terms of a physical exam.

6 I think how much you uncover from

7 somebody who is not really willing to go in, I

8 think, on a physical exam is somewhat limited. And

9 perhaps, depending on the questions that you

10 successfully ask them on the telephone interview,

11 you may be able to get a lot of the meat of what you

12 really want.


14 pursue it very far, it's just a thought in terms of

15 trying to deal with the non-respondents and

16 respondents and how comparable they are.

17 I wanted to ask Ms. Eydt about the OMB,

18 whether you feel that OMB should be involved earlier

19 and proceed or whether this experience and your

20 words to the agencies is sufficient to make sure

21 that the coordinating board of the other mechanisms

22 are utilized so that you have less problem when it

23 gets to you?

24 MS. EYDT: Well, like I said in my

25 remarks, at a minimum, I think, we should be


1 involved within the 60-day period when the

2 instrument is out for public comment.

3 We have also, in the past, been involved

4 with coordinating groups, we've been involved in the

5 formulation of RFP and have provided conceptual

6 clearances on packages before we see instruments.

7 So it is possible that we could become

8 more involved and it might be a very good idea.


10 question I have for Dr. Kang and for you, you made,

11 obviously, a very strong point about trying to get

12 consistency and using the CDC and the Iowa.

13 Yesterday, we did hear the case

14 definition that CDC has developed for the specific

15 Pennsylvania Study and I would like to know whether

16 the questions in your survey now will enable you to

17 apply that case definition to your survey instrument

18 when you get it?

19 DR. KANG: I don't have detail

20 information on their criteria for cases so I can't

21 comment on it. But what I heard, yes, the results

22 of our survey can be applied to define the cases.

23 MS. EYDT: I would say that he has a

24 relatively rich listing of symptoms so hopefully

25 that will help that type of comparison.


1 But I will admit we did not consider the

2 Pennsylvania clusters when we did our review and it

3 probably would have been of great interest. And I

4 will investigate whether, at OMB, we reviewed that

5 project and if it was appropriate for our review

6 earlier on.


8 we all need to take a good hard look at that case

9 definition that's come out of the Pennsylvania and

10 see whether our other studies will be able to come

11 up with maybe a phone questionnaire; if it's not

12 covered by the other, you might be able to take a

13 look at that again.

14 Okay, that's all I have.

15 MS. JOELLENBECK: I have a question for

16 Dr. Kang.

17 You had pointed out the strength of the

18 study being its providing, perhaps, the best

19 estimate of prevalence. But it is relying on self-

20 report. And you've built into it going back and

21 looking at medical records and having physical exams

22 for 2,000.

23 If that were done earlier in the study,

24 you might have a better sense of how reliable the

25 self-reports were, you might then not have to mail


1 out to 20,000 if you were to find that the self-

2 reports weren't as valuable as the information you

3 got from the exams and the record review.

4 DR. KANG: Uh-huh.

5 MS. JOELLENBECK: Is that something that

6 occurred to you in the planning process?

7 DR. KANG: I don't think I'm getting the

8 part of your question.

9 MS. JOELLENBECK: Well, doing the -- of

10 the 2,000 that you plan to review records and do

11 physical exams --

12 DR. KANG: Right.

13 MS. JOELLENBECK: -- and your reason for

14 doing that is to see how reliable the self-reports

15 are, is that --

16 DR. KANG: Yeah, okay.

17 MS. JOELLENBECK: So, if you were to do

18 that up-front and to find that, in fact, self-

19 reports were not as useful --

20 DR. KANG: But, see, the symptoms, we

21 don't have any core standard to compare against.

22 The only thing we can validate is the medical

23 conditions and the hospitalizations.

24 So for symptoms, we don't have anything

25 to compare.


1 MS. JOELLENBECK: So, no matter what,

2 you have rely on the self-report?

3 DR. KANG: Yes.

4 MS. JOELLENBECK: Were the power

5 estimates based upon a 70-percent response rate or a

6 40 to 60 response rate?

7 DR. KANG: Well, I just presented

8 numbers, the number of respondents that you need to

9 have for the statistical power if, in fact, the

10 difference is twofold or 50 percent. So it doesn't

11 really matter what your response rate is going to

12 be, that will be final number.

13 But we hope that the response rate will

14 be about 60 percent.

15 MS. JOELLENBECK: One final question.

16 Was there some kind of pilot testing of

17 the questionnaire with a veteran group or a group

18 that might be comparable.

19 DR. KANG: We did pilot testing using

20 less than ten people. We did pilot testing, yes.

21 COMMITTEE CHAIR LASHOF: You are shaking

22 your head, Dr. Kipen.

23 DR. KIPEN: I am. I find a pilot test

24 with less than ten people, while I recognize it

25 complies with regulations, to be of limited general


1 reliability.


3 there should have been further pilot testing before

4 it goes out?

5 DR. KIPEN: I didn't actually consider

6 it in this case, I was kind of shaking my head

7 generically there. But, in general, I am a big fan

8 of pilot testing.

9 MS. EYDT: I would like to add that we

10 also generate like -- we do always like good pilot

11 testing and there is no regulation that says you're

12 supposed to do pilot test of less than ten, it's

13 because they don't want to come to OMB, they do

14 pilot tests of less than ten.


16 Bureaucracy does have its problems.

17 MS. EYDT: That's an unintended effect,

18 I'm afraid.


20 John?

21 DR. BALDESCHWIELER: I'm curious as to

22 how you actually get information, for example, on

23 chemical sensitivity by asking questions.

24 Could you give me an example of the kind

25 of question that might appear on such a survey to


1 get information on chemical sensitivity? Either

2 Howard or Han?

3 DR. KANG: May I defer to Dr. Kipen?

4 DR. KIPEN: Well, there is one question

5 on it which is taken -- actually borrowed -- from a

6 survey that's been going on randomly in California

7 sponsored by the CDC for a number of years and it

8 simply asks, "Do you consider yourself unusually

9 sensitive to..." and it lists about four or five

10 things, perfumes, scented products, exhaust fumes,

11 and the like. And so that's one question that is on

12 the survey.

13 Our estimates from previous use of

14 questions like that would suggest that as many as 15

15 percent of randomly selected Californians will say

16 yes to that. And --


18 Californians typical?

19 DR. KIPEN: No, they are just --

20 And work we did a few years ago in

21 preparing an application to the VA for a Persian

22 Gulf center suggested as many as 40 percent of

23 Registered veterans will say yes to that.

24 So that would tend to get -- and that

25 was one of the reasons that I, among others on the


1 committee, thought that it would be nice to have

2 some follow-up questions which might be a little

3 less sensitive and a little more specific. Those

4 are not on the questionnaire.

5 There is some debate in the -- well,

6 there's no consensus and there's no gold standard

7 for what one would ask. Among the proposed

8 questions would be to ask people if they've changed

9 their lifestyle because of perceived increase in

10 sensitivity. And many fewer people, many people who

11 say, "Yes, I think I'm unusually sensitive" will not

12 say, "I've changed my shopping habits" or my eating

13 habits or my dressing habits and the like.

14 And so that's one of the approaches that

15 can be taken.


17 MR. BROWN: A question for Dr. Kang.

18 We heard yesterday from some, during

19 testimony, about the concern of some veterans or

20 people who are related to veterans that the health

21 effects -- that there seems to be little work

22 looking at health effects in family members other

23 than immediate spouse and children.

24 And I'm wondering, for the purposes of

25 your survey, when you're looking at health effects


1 in the National Survey -- well, you haven't gotten

2 to that -- when you're looking at health effects in

3 family members, how exactly do you define --

4 DR. KANG: It's limited to a spouse and

5 children.

6 MR. BROWN: So it wouldn't look at

7 roommates, siblings? Okay.


9 questions?

10 If not, I want to thank you very much.

11 We are running a little behind but we

12 will take a break now and, Dr. Kang, I understand

13 you took the red-eye to come and we do appreciate

14 that very much.

15 And thank you.

16 We will resume again at 10:10; we'll

17 take a 15-minute break.

18 (Whereupon, a brief recess was taken.)


20 please take their seats; we would like to get

21 started. We are a little behind schedule and we'd

22 like to move ahead as expeditiously as we can.

23 Thank you.

24 We will now move to a discussion of a

25 group of studies being carried out at the San Diego


1 Naval Research Center.

2 Dr. Gray -- well, I guess I don't know

3 everybody else, Kevin Kaiser, Bruce Coate, and David

4 Cowan, all from that unit. And, Dr. Gray, you're

5 going to kick it off and introduce your colleagues.

6 DR. GRAY: Thank you, Dr. Lashof.

7 Yes, I would like to introduce some of

8 my colleagues. To my left, are Dr. Kathy Araneta,

9 and Dr. Larry Dlugosz, who are leading some of the

10 follow-on studies from initial data studies which

11 you will hear about today.

12 We also have members of our Scientific

13 Advisory Panel. We have Dr. Warren Winkelstein, Dr.

14 Hal Morgenstern, and Dr. Phil Carter here today, as

15 well.


17 methodology will be for you to proceed through all

18 of your discussion of all of the studies, then we

19 will ask Dr. Winkelstein, Dr. Morgenstern, and Dr.

20 Carter to come forward. And then we will open it to

21 the panel; okay?

22 DR. GRAY: Very good.


24 DR. GRAY: What I would like to do today

25 is basically give you an overview of our seven


1 epidemiologic studies and talk about some of the

2 methodological problems that we've encountered.

3 (Slide change)

4 These studies are designed by a number

5 of investigators at a number of institutions. Here,

6 you see some of the institutions. In addition to

7 these qualifications, a number of our investigators

8 are veterans of the Persian Gulf War and four of our

9 investigators designed "hot pursuit", as the NIH

10 have called them, studies of various morbidity risks

11 among the Gulf War veterans.

12 (Slide change)

13 Here you see one of our recent

14 investigator meetings, collaborator meetings, where

15 our Scientific Advisors met in San Diego. And we've

16 had several of these in addition to the Scientific

17 Advisory Panel review which you'll see detailed in

18 the handout that I passed to the Committee earlier

19 today.

20 Additionally, our studies have been

21 reviewed by the Defense Science Board, the Institute

22 of Medicine, and the Armed Forces Epidemiological

23 Board.

24 (Slide change)

25 Our objectives in these studies are to


1 characterize the prevalence of illnesses, symptoms,

2 infertility, and adverse pregnancy outcomes among

3 military personnel before the Persian Gulf War.

4 Additionally, we wish to determine if military

5 service in the Persian Gulf War was associated with

6 illness.

7 (Slide change)

8 For the purposes of these studies, we

9 define military personnel as deployed to the Persian

10 Gulf War if they were in the geographical area

11 depicted here anytime from the 2nd of August, 1990,

12 when Iraq invaded Kuwait, until 31 July, 1991,

13 nearly a year later. This map and each stage

14 reflect data which were provided by the various U.S.

15 military services and recorded by the Defense

16 Management Data Center in Monterey, California.

17 This data base is the standard from which most

18 studies of veterans of the Persian Gulf War are

19 performed.

20 (Slide change)

21 In addition to telling us which persons

22 were veterans of the Persian Gulf War, the same

23 organization, Defense Data Management Center, also

24 provided data regarding which military personnel

25 were not so deployed. We distinguish these two


1 groups by the terms "deployed" to the Gulf War and

2 "not-deployed."

3 (Slide change)

4 In developing these studies, we

5 encountered a number of tough issues. It's

6 difficult to look for a disease which is not well-

7 defined. The fact that many of the former service

8 members have left the military has increased the

9 complexities of our research. Choosing appropriate

10 comparison groups has been difficult and, with the

11 many unusual exposures our military personnel

12 encounter, there is a chance that we will find

13 associations of exposure and disease which may not

14 be real. So we're concerned regarding how best to

15 validate our findings.

16 (Slide change)

17 The first of these issues involves

18 defining appropriate outcomes to study. And it's

19 been a problem for us from the beginning.

20 If a new condition has arisen with

21 numerous objective manifestations such as symptoms

22 of fatigue and joint pain and there are no

23 biological tests available to detect it, how do we

24 effectively screen for it?

25 In a similar fashion, if veterans of the


1 Persian Gulf War are sicker than their peers, and

2 the illnesses have not been defined, how will you

3 compare their hospitalization experiences?

4 (Slide change)

5 We decided to take a three-pronged

6 exploratory approach designing studies to gather

7 information regarding symptoms, hospitalizations,

8 and birth defects. Necessarily, these studies are

9 broad and we evaluate many outcomes. The reason

10 that, after examining these numerous outcomes, if we

11 find a pattern of certain veterans with certain

12 exposures are in higher risk, then we would target

13 these individuals with more comprehensive research.

14 (Slide change)

15 Regarding symptoms, we performed a

16 cross-sectional survey of Navy construction workers,

17 or Seabees, who were on active duty at the time of

18 the Gulf War and have remained on active duty.

19 Seabees were some of the first personnel

20 to report unusual symptoms which they attributed to

21 the Persian Gulf War. We've collected questionnaire

22 data, have performed physiological testing, and

23 collected clinical specimens from 1500 service

24 persons.

25 (Slide change)


1 Regarding hospitalizations, we performed

2 a retrospective cohort study of the Department of

3 Defense hospitalization experience of 1.2 million

4 service persons who were on active duty at the time

5 of the Gulf War. These analyses involve all

6 Department of Defense hospitals throughout the

7 world. Additionally, we've examined the birth

8 outcomes from these 1.2 million persons and their

9 spouses.

10 These three studies are well underway

11 and you'll hear about some of their results this

12 morning.

13 (Slide change)

14 A limitation of these initial three

15 studies is that we were only capturing data from

16 service personnel who remained on active duty. We

17 began our first survey work in September, 1994 and

18 you can see that a significant portion of the

19 original 697,000 military persons who were deployed

20 to the Persian Gulf War were no longer on active

21 duty at that time. And, of course, this percentage

22 continues to increase so our studies might miss

23 important findings among former military personnel

24 who left the military service.

25 (Slide change)


1 Hence, we've designed four other studies

2 to examine former military personnel, as well as

3 those who remained on active duty. These new

4 studies are in their early stages of data

5 collection. They include a large mail survey of

6 17,000 Seabees, analyses of civilian hospitalization

7 experiences focusing on California, a study of birth

8 defect registries maintained by seven states, and a

9 large mail and telephone reproductive survey of

10 16,000 couples.

11 (Slide change)

12 Our third issue area involved selecting

13 appropriate comparison groups from the service

14 personnel deployed to the Persian Gulf War. We're

15 concerned that deployed personnel might be healthier

16 than personnel who were not selected for deployment.

17 Sometimes we may avoid this potential bias by

18 examining outcomes only among the deployed group.

19 In other studies, we adjust for potential population

20 differences.

21 (Slide change)

22 We're concerned that news reports may

23 influence the way deployed personnel respond to

24 questions. We realize that self-reported data, such

25 as symptoms, are more likely to be influenced by


1 such recall bias than would measurable or objective

2 outcomes like a serologic test.

3 (Slide change)

4 However, it may be useful to study

5 softer outcomes, such as symptom complexes, with

6 serologic tests in order to evaluate various

7 hypotheses. For instance, once a serologic

8 screening test for leishmaniasis has been developed,

9 this test might be used to compare the most

10 symptomatic Seabees with those who have fewer

11 symptoms.

12 While we're not altogether satisfied

13 that we have solved the issue of recall bias, we

14 have designed a number of checks in our studies and

15 we are aware of this important potential problem.

16 I think we can look at some of the -- at

17 least roughly -- the CDC's proposed case definitions

18 through some of our initial data.

19 A final issue area involves the

20 question: How good are our data? How do we know if

21 our findings for association or lack of association

22 are real?

23 (Slide change)

24 By performing a number of studies, some

25 of which overlap, we may compare one study against


1 another. We can do this among several Department of

2 Defense studies and also compare the Department of

3 Defense studies with studies from other federal and

4 civilian institutions.

5 Regarding the quality of our data, we've

6 learned about the reliability of our surveys by

7 administering them to a portion of volunteers more

8 than one time. We've also validated some

9 questionnaire responses by reviewing service and

10 health records.

11 (Slide change)

12 These studies have led us to the

13 development of a bibliography of over 17,000

14 references which we've uploaded to the Gulf Link in

15 the Web sites listed here. Also, we benefit from

16 the input from various agencies including the DOD

17 Persian Gulf investigating team headed by Colonel

18 Koningsberg.

19 (Slide change)

20 In closing, I'd like to emphasize that

21 we're very concerned about the health of Persian

22 Gulf War veterans and their families. The

23 Department of Defense studies reflect multiple

24 research approaches which have received intensive

25 external scientific review. These studies, coupled


1 with studies from other federal and civilian

2 academic researchers will give us an excellent

3 picture of the morbidity of veterans of the Persian

4 Gulf War.

5 Dr. Kevin -- or Kevin Kaiser -- he's not

6 a doctor yet but he's working on it -- will talk a

7 little bit about our Seabee study.

8 MR. KAISER: Good morning everyone.

9 This study was initiated to examine the

10 possible associations between military service in

11 the Persian Gulf War and morbidity. Our emphases

12 were on symptoms and exposures that were not

13 captured by existing databases.

14 (Slide change)

15 We selected regular active duty Navy

16 construction workers, or Seabees, as our study

17 population for a number of reasons. Number one,

18 Reserve Seabees have reported a higher prevalence of

19 morbidity. Number two, regular active duty Seabees

20 have the same jobs and exposures as the Reserve

21 Seabees. Third, active duty Seabees are

22 concentrated in either California or Mississippi

23 when not deployed overseas and, each year, active

24 duty Seabees are deployed overseas for approximately

25 six months and, hence, experience frequent


1 deployment-related exposures.

2 Regular active duty Seabees are less

3 affected by civilian occupational experiences than

4 are Reserve Seabees who serve in the military on a

5 part-time basis and are exposed to other occupation-

6 related agents.

7 (Slide change)

8 During the period from September, 1994

9 through June of 1995, we made six trips to two

10 Seabee centers in California and Mississippi. We

11 enrolled Seabees who had remained on active duty

12 since the time of the Persian Gulf War. Both Seabee

13 military personnel who had been deployed to the

14 Persian Gulf War and other military personnel from

15 the same era who did not deploy to the Gulf War were

16 permitted to participate.

17 (Slide change)

18 Our cross-sectional survey collected

19 information regarding demographics, symptoms,

20 chronic diseases, hospitalizations, and birth

21 outcomes. This survey instrument was designed to

22 screen for chronic fatigue syndrome, post-traumatic

23 stress disorder, and five psychological symptom

24 dimensions from the Hopkins Psychological Symptom

25 Dimension Survey. This is an instrument to quantify


1 psychological status profiles.

2 And, for each Hopkins Psychological

3 Symptom Dimension, an average score for each

4 individual was calculated for the five dimensions.

5 And these dimensions are somatization, obsessive-

6 compulsive disorder, interpersonal sensitivity,

7 anxiety, and depression. Birth outcomes and

8 hospitalizations were assessed from July, 1990 and

9 blood and urine specimens were collected from each

10 study subject during our site visits for later

11 analyses.

12 (Slide change)

13 We also tested the reliability of the

14 study instrument by retesting a random sample of 260

15 individuals who are known to be available for

16 retesting at the time of our next study visit.

17 Similarly, we randomly sampled another 150

18 individuals out of this initial 260 to perform

19 validated checks by comparing selected survey

20 responses to medical and service records.

21 Due to low participation by some

22 commands, we also looked at the potential selection

23 bias by comparing volunteers and non-volunteers with

24 regard to demographics and Gulf War status.

25 (Slide change)


1 Handgrip strength was assessed by

2 calculating an average of three successive efforts

3 using a hand-held dynamometer.

4 (Slide change)

5 Lung function was measured as the mean

6 of two efforts and quantified according to

7 percentage predicted of the forced expiratory volume

8 in one second over the forced vital capacity.

9 (Slide change)

10 We enrolled a total of 1497 subjects

11 with 527 deployed to the Gulf War and 970 non-

12 deployed. After adjusting for availability during

13 our site visits, we estimate the participation

14 ranged from 38.6 to 61.5 percent of eligible

15 personnel.

16 The demographic characteristics of

17 participants and non-participants were very similar.

18 (Slide change)

19 Demographically, the two groups of

20 participating veterans was similar with respect to

21 race, marital status, height, and weight, but

22 differed in that deployed veterans were slightly

23 younger and more individuals whose highest degree of

24 attainment was a high school diploma.

25 (Slide change)


1 Veterans who had deployed to the Persian

2 Gulf War had higher scores on all five dimension

3 scales of the Hopkins Psychological Symptom Profile.

4 However, their mean scores were low compared to the

5 published mean scores of patients clinically

6 diagnosed with depression or anxiety disorders.

7 (Slide change)

8 Regarding their physical measurements,

9 individuals deployed to the Gulf War were no

10 different from the non-deployed for both handgrip

11 strength and spirometry.

12 (Slide change)

13 These are the eight most prevalent self-

14 reported symptoms in our study population.

15 Regarding these symptoms expressed for one month or

16 more since July, 1990, those deployed to the Gulf

17 War reported significantly higher prevalences for 35

18 of the 41 symptoms queried.

19 (Slide change)

20 However, a large portion of those

21 participating in this survey reported no symptoms at

22 all. This result is not surprising since the study

23 population consisted of active duty Seabees who are

24 expected to be in good physical condition in order

25 to accomplish the tasks required by their jobs.


1 (Slide change)

2 Additionally, it was noteworthy that,

3 among those deployed personnel who complained of

4 muscle weakness for one month or more since July,

5 1990, there were no major differences in handgrip

6 strength testing compared to deployed personnel who

7 did not report muscle weakness.

8 (Slide change)

9 Similarly, among those deployed

10 personnel reporting shortness of breath for one

11 month or more since July, 1990, there was no marked

12 difference in lung function by spirometry compared

13 to those who did not report shortness of breath.

14 (Slide change)

15 Subjects were asked about pregnancy

16 information on all of their partners and responses

17 were divided into two groups, those being live

18 births and other non-live birth outcomes, including

19 still birth, miscarriage, ectopic pregnancy, and

20 elective induced abortion. The results show no

21 difference between the groups regarding birth

22 outcomes with results derived from 590 birth events

23 in those deployed to the Gulf War and 1364 birth

24 events for those not deployed to the Gulf.

25 (Slide change)


1 Individuals were also asked about their

2 number of hospital visits since July, 1990.

3 However, unlike the results from birth outcomes, the

4 two groups were different with the deployed Gulf War

5 veterans reporting more hospitalizations than the

6 non-deployed group.

7 (Slide change)

8 Using several strategies, including

9 reviewing past investigations, multivariate logistic

10 regression, and univariate analyses, we designed

11 three diagnostic hypotheses each composed of a

12 unique symptom pattern. From our questionnaire, we

13 also derived similar screening tools for post-

14 traumatic stress disorder and chronic fatigue

15 syndrome.

16 Not one of the 1497 study subjects

17 screened for chronic fatigue syndrome; however, we

18 did find that personnel deployed to the Persian Gulf

19 War were more likely to screen for both post-

20 traumatic stress disorder and for the three

21 exploratory symptom patterns we designed.

22 (Slide change)

23 Gulf War veterans were more likely to

24 self-report exposures such as the eight most

25 prevalent shown here. However, examining various


1 exposures in a myriad of univariate and multivariate

2 models, more exposures were consistently associated

3 with the prior-mentioned study outcomes, which were

4 chronic fatigue, post-traumatic stress disorder, and

5 the three hypothesized symptom patterns.

6 (Slide change)

7 As with any study, there is some

8 limitations that may influence the results such as

9 recall bias involved in trying to answer questions

10 regarding symptom expression, health history, and

11 psychological status. Another factor was incomplete

12 participation in some commands which was influenced

13 by a multitude of factors.

14 However, we feel that selection bias was

15 not a significant factor in our analyses based on

16 the comparisons of those who participated and those

17 who did not.

18 Only currently active duty individuals

19 were studied so those Seabees who have separated

20 from the military since the Gulf War would be unable

21 to participate which possibly could influence the

22 results.

23 (Slide change)

24 In summary, those deployed to the Gulf

25 War had higher self-reported prevalences for 35 of


1 the 41 symptoms. Although, when specific symptoms

2 were compared with physical measurements, there was

3 no difference. Those deployed to the Gulf War had

4 higher average scores than the non-deployed for all

5 five of the Hopkins Symptom Dimensions, however it

6 is not clear whether this is reflective a pre-Gulf

7 War or a post-Gulf War condition.

8 Third, we found no major difference in

9 physical performance or self-reported birth outcomes

10 between the groups. And fourth, no single exposure

11 was consistently associated with any of our study

12 outcomes.

13 (Slide change)

14 Future analyses will focus on possible

15 latent effects of Gulf War service by conducting a

16 more comprehensive survey of 17,000 Seabees

17 including both active duty and Reservist

18 individuals. This will also complement our study of

19 solely active duty military.

20 Our stored pre-Gulf War and post-Gulf

21 War sera will also be used to test hypotheses

22 regarding infectious agents as they emerge.


24 Bruce Coate will express some of our hospital study

25 data.


1 MR. COATE: I'm going to talk about the

2 results of our study on morbidity of regular active

3 duty Gulf War veterans as measured by

4 hospitalization discharges.

5 (Slide change)

6 The objectives of our study were to

7 describe the differences in hospitalization rates

8 between those who deployed to the Persian Gulf and

9 those who did not deploy.

10 Secondly, we were seeking to identify

11 specific disease categories which may warrant

12 further investigation so our analyses were mainly

13 exploratory in nature rather than testing any

14 specific hypothesis.

15 (Slide change)

16 Our study design is a retrospective

17 cohort study covering the period from August 1st,

18 1991 through September 30th of 1993. We chose these

19 dates because, as of August of 1991, over 95 percent

20 of the deployed personnel had returned from the Gulf

21 and September, 1993 is the latest date for which

22 hospitalization data were available.

23 (Slide change)

24 Our study population consisted of two

25 groups, the deployed and non-deployed, the deployed


1 being those regular active duty service members who

2 were in the Persian Gulf anytime between August 1st,

3 1990 and July 31st of 1991, while the non-deployed

4 were active duty service members who did not deploy

5 during this time and who were on active duty as of

6 September 30th of 1990.

7 So this sample includes personnel from

8 all branches of the military but does not include

9 Reservists or National Guard members, who accounted

10 for about 100,000 of the deployed troops.

11 (Slide change)

12 This is just a comparison of the two

13 groups that affected a few variables. We have about

14 99 percent of the deployed personnel in our sample.

15 And, with respect to the non-deployed, we have a 50

16 percent random sample stratified by branch of

17 service so we have obtained follow-up information on

18 700,000 of the 1.4 million non-deployed troops.

19 We can see that there is a larger

20 percentage of males among the deployed personnel and

21 the deployed personnel were somewhat younger than

22 the non-deployed.

23 (Slide change)

24 Our data was obtained from the

25 Department of Defense Manpower Data Center and


1 hospitalization files contained discharge records

2 covering the period from 1989 through 1993 and each

3 discharge could contain up to eight separate

4 diagnoses. The diagnoses were coded according to

5 the International Classification of Diseases and

6 Injuries, Ninth Revision.

7 (Slide change)

8 The outcome variable in each of our

9 analyses was essentially the same, mainly the

10 occurrence or non-occurrence of a hospital

11 discharge. We divided our follow-up period into

12 three eras, it would be the last five months of

13 1991, all 1992, and the first nine months of 1993.

14 Within these time frames, we examined an all-cause

15 hospitalization outcome as well as 14 category-

16 specific hospitalizations.

17 (Slide change)

18 This is a list of the broad disease

19 categories we examined, diseases associated with

20 pregnancies, congenital anomalies, and perinatal

21 conditions are not included in our list of outcomes,

22 as these were dealt with in Dr. Cowan's study.

23 (Slide change)

24 This is a graph of the crude all-cause

25 hospitalization rate from October, 1988 through


1 September of 1993 for the deployed and non-deployed.

2 The crude rate per 100,000 is on the vertical axis

3 and the month is on the horizontal axis. We divided

4 the time scale into three periods, a pre-War period,

5 the Gulf War era, and the post-War period.

6 During the pre-War era, the rate is

7 generally higher among the non-deployed with this

8 difference becoming accentuated in the months just

9 prior to deployment. These are not too surprising

10 since the personnel who have recently been

11 hospitalized are going to be less likely to be

12 eligible for deployment overseas.

13 During the second period when the troops

14 were in the Gulf, the difference in hospitalization

15 rates became even more pronounced reflecting, among

16 other factors, low battle casualty rates, absence of

17 serious disease outbreaks, and deferral of elective

18 surgeries.

19 During the post-War era, the rates were

20 still high among the non-deployed but the difference

21 is somewhat less pronounced than it was earlier.

22 (Slide change)

23 This slide shows the post-War crude

24 hospitalization rates per 1,000 -- between the two

25 groups for each of the 14 specific disease


1 categories we examined. The crude hospitalization

2 rates were generally higher for the non-deployed

3 with the exceptions of injury and poisoning, which

4 is C14 over there, and C5, which is mental

5 disorders.

6 We used two basic statistical

7 methodologies to examine our data, standardized

8 rates and multivariate logistic regression.

9 (Slide change)

10 This table just shows the various models

11 which we fit for various time periods. And we were

12 only concerned with logistical regression results,

13 the other, the more exploratory time analyses, and

14 all the stuff I'm going to talk about today will

15 just be logistic regression results and rate ratios.

16 (Slide change)

17 Our directly standardized rates were

18 obtained by applying age- and sex-specific cohort

19 rates to the standard distribution which in this

20 case was the combined population of deployed and

21 non-deployed, and each subject could count for, at

22 most, one hospitalization per year per category.

23 (Slide change)

24 The standardized rate ratio of greater

25 than 1.0 would indicate that the deployed were at


1 increased risk for hospitalization relative to the

2 non-deployed. This occurred only for the 1992 time

3 period but the effect is still quite small; the rate

4 ratio of 1.02 was also a little bit higher in 1993

5 but that was not statistically significant.

6 (Slide change)

7 This is a list of the variables which we

8 included in our multivariate logistic regression

9 models, an indicator variable for deployment status

10 which was fixed in each model while the other

11 variables were kept in the model based on a

12 backwards elimination procedure. And we categorized

13 all our continuous variables into discrete levels.

14 And we also attempted to adjust for pre-War health

15 status by including a pre-War hospitalization rate.

16 But this was dropped from our final models because

17 it has little effect on our odds ratios.

18 (Slide change)

19 These are the adjusted odds ratios for

20 the all-cause hospitalization outcome adjusted for

21 the factors listed in the previous slide. The odds

22 ratios are all quite close to 1.0 indicating that

23 the deployed were not at increased risk for

24 hospitalization relative to the non-deployed.

25 In addition to these all-cause outcomes,


1 we also fit models for each of the category's

2 specific outcomes. Of these models, only two were

3 significant, and these are listed in the next slide.

4 (Slide change)

5 In both instances, there's only a

6 moderate risk increased hospitalization for the

7 deployed but, because of the multiple models which

8 we set, numbering about 42, this could have just

9 occurred by chance.

10 (Slide change)

11 But we looked at these categories in a

12 little more detail. These are the four most

13 prevalent diagnoses for the mental disorder category

14 in 1992, these accounted for approximately 75

15 percent of all hospitalizations in this category.

16 The largest difference in rates is found

17 in the alcohol dependence diagnoses, followed by

18 adjustment reaction.

19 (Slide change)

20 Among the genitourinary diseases, the

21 highest rates occurred among females with deployed

22 women having higher rates of hospitalization for

23 pain and other symptoms from inflammatory diseases.

24 (Slide change)

25 The limitations in our study are that --


1 the most serious one probably is the loss to follow

2 up. By the end of our study period approximately 40

3 percent of the population have left the military.

4 We have no information on potential compounders such

5 as diet, weight, and smoking.

6 And we have looked at hospitalizations

7 in the military facilities so, if non-military

8 hospital use is different among deployed, this may

9 bias our results. But we feel this is unlikely to

10 occur since it is difficult for a regular active

11 duty service member to seek in-patient care outside

12 of the military system. And also, as mentioned

13 earlier, we didn't include any Reservists or

14 National Guard in our sample.

15 (Slide change)

16 The strengths of our study are its

17 perspective design, the factoring of information on

18 the large number of individuals from all branches of

19 the military and that we were able to work with a

20 large number of outcomes.

21 (Slide change)

22 In conclusion, when hospitalizations

23 among broad categories were examined using logistic

24 regression and age- and sex-adjusted rates, deployed

25 personnel were found to generally not be at


1 increased risk for hospitalization.

2 Two exceptions were the mental disorders

3 and diseases of the genitourinary system but these

4 results were significant for only one time period

5 and were not consistent the length of the entire

6 study period. We are planning on doing further

7 descriptive analyses of these two categories in the

8 near future.

9 Thank you.


11 presentation involves a similar approach; Dr. David

12 Cowan will discuss birth defects as occurred in the

13 DOD hospitalization system.

14 DR. COWAN: Thank you.

15 Please let me know if this is not loud

16 enough.

17 We've all seen the Life magazine

18 article with these tragic stories and photographs,

19 yet individual case reports are not adequate to

20 establish if there is an association between Gulf

21 War service and the risk of birth defects.

22 (Slide change)

23 In order to assess risks, epidemiologic

24 studies utilizing appropriate comparison groups are

25 required. To that end, it is important that we do


1 our very best to measure the risk of events

2 occurring in these populations in determining if the

3 risk of defects occurring among children of Gulf War

4 veterans is actually higher than the risk among

5 military members who did not serve in the Gulf.

6 In order to contribute to the

7 understanding of this situation, we have initiated

8 this epidemiologic study of the risk of birth

9 defects among children of Gulf War veterans and

10 other military personnel.

11 (Slide change)

12 Hundreds of thousands of men and women,

13 soldiers, sailors, airmen, and marines serve in our

14 armed forces each year. Many of them have children

15 while they or their spouses are on active duty and

16 these children are often delivered in a military

17 hospital.

18 For our study population, we identified

19 all the active duty Gulf War veterans and a sample

20 of active duty non-Gulf War veterans.

21 (Slide change)

22 We conducted a historical cohort study

23 in which the individuals were identified based on

24 their deployment status and were followed over time

25 to identify live births and then defects occurring


1 in military hospitals. This is a graphic

2 representation of our study.

3 (Slide change)

4 In this study, we are not evaluating any

5 specific environmental exposures. Rather, we are

6 comparing the risks of defects between those who

7 served in the theater of operations and those who

8 did not. In addition, we are examining the data to

9 see if there is an association between duration of

10 time in the theater and the risk of birth defects.

11 The data are stratified and separate

12 analyses presented for active duty women and wives

13 of active duty men.

14 (Slide change)

15 We accessed an existing administrative

16 database to identify live births to active duty

17 personnel occurring in military hospitals. Children

18 with estimated dates of conception after the return

19 of the sponsor from the Gulf or after the 1st of

20 January, 1991 for non-deployed members and with

21 dates of birth before the 1st of October, 1992 were

22 eligible for study inclusion.

23 (Slide change)

24 This is the number of births occurring

25 to our study subjects. Note that, in the early


1 months, the non-deployed military personnel had

2 higher numbers of babies but that, as they re-

3 deployed back to the United States, the Gulf War

4 veterans rapidly caught up.

5 (Slide change)

6 Medical diagnostic data were acquired

7 from an administrative database which is routinely

8 created for each in-patient in military hospitals.

9 Note that these diagnoses are from the birth

10 hospitalization only. ICD-9 codes are used with up

11 to eight different diagnoses captured and the

12 diagnoses are up to five digits in length.

13 (Slide change)

14 We looked at two non-birth-defect

15 outcomes, live birth rates and the sex ratio of

16 newborns. For definitions of the birth defects, we

17 used the coding manual from the Metropolitan Atlanta

18 Congenital Defects Program.

19 (Slide change)

20 The MACDP had been in existence for

21 several years and was developed for the Centers for

22 Disease Control. We utilized their diagnostic codes

23 in order to have an externally validated coding

24 scheme. The MACDP uses the ICD-9 codes which gave

25 compatibility with the military coding system.


1 There are some important points I'd like

2 to make. The MACDP includes other codes besides

3 just the malformation codes such as neoplasms and

4 hereditary diseases. In the Atlanta Program, some

5 minor cases of certain defects are excluded based on

6 medical records review. Because our study did not

7 include a medical records review, minor cases of

8 certain defects were retained in our data set.

9 (Slide change)

10 For our statistical analyses, we used

11 Chi square to assess linear trend, relative risks to

12 assess univariate associations, Mantel-Haenzel

13 Summary relative risks to assess potential

14 confounders, and logistic regression to develop

15 adjusted odds ratios.

16 Note that all comparisons I will present

17 are statistically significant unless otherwise

18 noted.

19 (Slide change)

20 Here we see our study population broken

21 out by Gulf War service and sex. These are active

22 duty men, Gulf War veterans and non-Gulf War

23 veterans, active duty women, veterans and non-Gulf

24 War veterans.

25 We ascertained attrition by identifying


1 those in the original cohort that were still

2 participating on active duty at the close of the

3 study period. Note that the attrition rate is quite

4 similar for both comparison groups.

5 Excuse me -- this is the survival rate,

6 the percent remaining on active duty.

7 (Slide change)

8 Now we move to the next step of the

9 results, the numbers of live births occurring in

10 each cohort.

11 (Slide change)

12 On this slide we see the total number of

13 births by cohorts and the accrued live birth rate.

14 Note that this is not a time-based rate, but numbers

15 only, a proportion.

16 Even though there are significant

17 differences between the cohorts and the live birth

18 rate, the numbers are -- the proportions are very

19 similar. The male to female ratio of the newborns

20 is shown and there are no significant differences

21 between these ratios for the comparison groups for

22 either men or women.

23 (Slide change)

24 Here we see the numbers of children with

25 birth defects born to each cohort.


1 (Slide change)

2 This slide shows the percent of live

3 births with the coded defect for each of the

4 comparison groups. At the bottom of the slide is

5 the crude relative risk for Gulf War veterans

6 compared to non-veterans, 1.02 for active duty men

7 and 1.13 for active duty women. Please note that

8 there are no statistically significant differences

9 in these comparisons.

10 (Slide change)

11 On this slide, we examine the data for

12 an association between duration of time in the Gulf

13 theater of operations for men and women. None of

14 the individual time periods are significantly

15 associated with increased risk. The competence

16 intervals for all of the odds ratios include the

17 value of 1.0.

18 Furthermore, there was no significant

19 linear trend for increasing risk with increasing

20 duration of time in the theater.

21 (Slide change)

22 Finally, we conducted multivariate

23 analyses controlling for several potential

24 confounding factors. The final model for both men

25 and women controlled the race, ethnicity of the


1 sponsor, age of the mother at time of birth, the

2 branch of service, and the rank of the sponsor. We

3 found no significant association between Gulf War

4 service, yes or no, for men or for women.

5 None of the individual time periods was

6 significantly associated with an increased risk of

7 birth defect. And, when the duration of time spent

8 in the theater was entered as a continuous variable,

9 it was not a significant predictor.

10 (Slide change)

11 There are limitations to this study

12 which restrict the interpretation. We assessed only

13 live births of active duty personnel occurring in

14 military hospitals and we only captured data from

15 routinely-generated administrative sources. Because

16 our coding system is based on that used by an active

17 case-finding program, there is limited

18 generalizability beyond our comparison groups.

19 For all of these reasons, comparisons

20 with published rates of birth defects from civilian

21 populations may not be appropriate.

22 (Slide change)

23 This study also has important strengths

24 including a large number of observations so that we

25 had good statistical power to detect differences


1 that may have existed. We had the exposure and

2 birth outcome data for both men and women service

3 members. All births occurred shortly after the war,

4 presumably when any affects of wartime exposures

5 would have been most pronounced.

6 Our multivariate analyses controlled for

7 potential confounders including surrogate measures

8 of socioeconomic status. Our attrition rates are

9 similar between groups and we utilized the same

10 exposure and outcome data sources for deployed and

11 non-deployed members.

12 (Slide change)

13 Finally, our conclusions are that we

14 found no significant association between Gulf War

15 service and the overall risk of birth defects for

16 children of women or men in our cohorts and we found

17 no significant dose response relationship between

18 duration of Gulf War service and the overall risk of

19 birth defects for children for men or women in our

20 study.

21 Thank you.


23 Gray, does that complete the studies that are

24 ongoing at this time?

25 DR. GRAY: Yes, Dr. Lashof.



2 think what I'd like to do at this point, advance Dr.

3 Winkelstein, Dr. Morgenstern, Dr. Carter to come

4 forward and if you'll just take seats in the front

5 and then we will have you all. So we will figure

6 out how we will arrange the seating when we get to

7 the question and answers.

8 Commander Gray and your group would take

9 seats in the front row and let Dr. Carter,

10 Morgenstern, and Winkelstein take the podium there

11 where they have the mikes. Then we will try to

12 rearrange things after the three of you finish so

13 that we can get everybody together in a way that at

14 least we can ask questions.

15 Thank you very much for coming and we'd

16 like you -- all of you have been involved in the

17 scientific review of the cases done at the Naval

18 Research unit, the San Diego unit, and we'd like to

19 hear your comments at this time.

20 Dr. Winkelstein, do you want to start?

21 DR. WINKELSTEIN: Thank you, madam

22 Chairman -- Chairperson.

23 My role was to act as a sort of advisor

24 at a meeting at which the protocols were presented

25 last February in San Diego. And the committee of


1 three, Dr. Baird, Dr. Morgenstern, and myself

2 produced a report as a result of that meeting and of

3 our examination of the protocols.

4 And I know you have copies of that

5 report in your briefing book.

6 Since that time, I was at San Diego for

7 sort of a review session in August and wrote a brief

8 memorandum as a result of the August meeting; and

9 that, too, you have in your briefing book.

10 I did not attend the APHA meeting at

11 which the results of the studies were presented so

12 that the first chance of seeing the results was

13 today. And I was pleased to see that some of the

14 suggestions that we had made in our report had been

15 implemented in the report of the findings.

16 I'd like to address very briefly some of

17 the issues of the Seabee study which was of

18 particular interest to me. I missed the statement

19 of the participation rate but I want to address that

20 problem first.

21 The Seabee study of symptoms is based on

22 a sample, a volunteer sample, obtained many years

23 after the actual exposure period and, at the time we

24 reviewed the work last February, the participation

25 rate was 50 percent. And I believe it ended up a


1 little bit better than 50 percent, although I'm not

2 sure what it was.

3 Could you tell me that?

4 MR. KAISER: (Inaudible comment from an

5 unmic'ed location.)

6 DR. WINKELSTEIN: So the final -- what

7 do you mean by the range?

8 MR. KAISER: (Inaudible from an unmic'ed

9 location.)

10 DR. WINKELSTEIN: So the final

11 participation rate was somewhere in the neighborhood

12 of 50 percent, as just reported between 38 and 65

13 percent, so roughly 50 percent.

14 And I think that is what is the major

15 problem with interpreting the findings. The

16 findings were, in general -- showed that there were

17 excess symptoms among deployed personnel but it's

18 difficult to interpret that when you're faced with a

19 50 percent participation rate and the 50 percent

20 were all volunteers.

21 So I don't really know how to interpret

22 that. As I said yesterday, in connection with the

23 Pennsylvania study, if deployment status and symptom

24 status is related to participation, then you have

25 the potential for bias and you can get either -- any


1 result that you get is always subject to question.

2 So that's really my main problem with

3 the Seabee study. I think that the fact that they

4 did a retest and that they did a record survey to

5 validate what they found, and even if they have

6 compared the characteristics of participants,

7 general characteristics of participants with non-

8 participants, we're still left with a problem of

9 interpretation.

10 Now the second Seabee study, which was

11 only mentioned today, I think we have a new protocol

12 in the briefing book and I was able to look at it

13 briefly yesterday and today. It's a revision of the

14 protocol that was presented in August.

15 And originally, in February, the mail

16 survey proposal was designed to deal with some of

17 the problems of the original Seabee survey. The new

18 protocol is not clear to me exactly what the long-

19 term objectives are but it includes at least mention

20 of the possibility of a follow-up, as long as 15

21 years. And that presents, it seems to me, immense

22 problems in terms of selection, follow-up, biases,

23 and costs.

24 And that's not very clearly described in

25 the protocol. So I would just draw your attention


1 to those problems.

2 That's all I have to say at this point.


4 you.

5 I think we will ask you to give the

6 comments and then we will have the questions.

7 Dr. Morgenstern?

8 DR. MORGENSTERN: I'm Hal Morgenstern,

9 Professor of Epidemiology at UCLA School of Public

10 Health and I was asked to participate as a member of

11 the Scientific Advisory Committee for the Naval

12 Health Research Center some time in January and

13 agreed to do so.

14 The idea was that we would review the

15 seven studies that were being conducted by the Naval

16 Health Research Center, as well as a few that were

17 being planned in different phases. And we met for a

18 two-day site visit in San Diego in February and, as

19 you know, we assembled a report over the next month

20 to month and a half which was issued to the

21 appropriate Defense Department -- Office of the

22 Defense Department in late March.

23 During the two-day site visit, we were

24 asked if we were interested in participating in

25 future ongoing efforts by the study team and, as I


1 recall, all three of us agreed, and I did, too; we

2 were planning to actually be there in August for

3 another meeting.

4 During the next couple of weeks,

5 sometime after the report was issued, I was sent a

6 message, I think it was a FAX, by Commander Gray

7 asking if I would be interested in participating

8 again in August, even though there were no funds to

9 support our time but there was opportunity to pay

10 for our expenses. And there was also a comment

11 about not wanting us to distribute our share of the

12 report that we wrote with anyone.

13 And so I was, of course, still

14 interested in participating and left a message with

15 Commander Gray, which he never answered. So I have

16 no knowledge of actually what was done after

17 February or March on any of these projects except

18 what I heard here this morning.

19 So it's difficult for me to comment on

20 the progress that's been made and how I would view

21 the studies in terms of opportunity for drawing

22 conclusions that were relevant to the mandate of

23 those studies.

24 We made a number of recommendations; I

25 was actually responsible for reviewing and writing


1 about the hospitalization studies. And one of the

2 things that concerned us -- two of the things that

3 concerned us at the two-day site visit, is first

4 that the protocols that were laid out to do those

5 studies were rather incomplete and were very

6 difficult for us to sort out what the members of the

7 team were trying to accomplish. But much of that

8 was supplemented at the site visit itself, when we

9 found out what the team was doing.

10 It appeared to me that the studies were

11 designed rather hastily initially, possibly in

12 response to some funding that came about very

13 quickly. I don't know the history of it too well, I

14 just know that this funding started in 1994.

15 It appeared that the first three studies

16 that you heard reported here today were initiated

17 and then the investigators apparently realized that

18 there were some potential problems, sources of bias

19 that would impede the ability to make causal

20 inferences from those studies.

21 So they initiated three other studies to

22 supplement those findings and make up for the bias,

23 so to speak.

24 I remember that my reaction to that, and

25 I believe it was the position of the entire advisory


1 panel of three, was that, rather than trying to do

2 three studies that were obviously limited for

3 drawing important inferences about this very

4 important topic, rather they should try to integrate

5 the concerns that they were doing in the

6 supplementary studies and do one study of each of

7 these topics and try to do it appropriately with

8 minimal bias by trying to study as properly as they

9 could the population at risk of interest.

10 My impression this morning is that that

11 may not have been done. It still appears that the

12 major limitations of all of the studies that we

13 pointed out are still major limitations. That's not

14 to suggest that the group, the study team, hasn't

15 advanced their ideas of how to do the studies and

16 doing them well; I suspect they've done quite a bit.

17 And certainly, by virtue of seeing the results here,

18 I can see that a great deal of work was done since I

19 was involved.

20 But nevertheless, I still think that

21 there will be a number of important questions that

22 will remain about the validity of the results that

23 we just saw and the conclusions that were drawn.

24 But again, it's hard for me to comment

25 on the specifics right now since all I've heard


1 since February were the results that you just saw.


3 much.

4 Dr. Carter?

5 DR. CARTER: Thank you, madam Chairman.

6 The Committee has my biography but, for

7 clarification and the benefit of the audience, I'd

8 like to take a minute to tell you why I'm here.

9 My expertise is not in epidemiology so I

10 shan't comment on the studies that you heard

11 presented.

12 I'm a full professor since 1982 of

13 microbiology and immunology at the College of

14 Veterinary Medicine in North Carolina State

15 University. In the State of North Carolina, we have

16 two major military installations, Fort Bragg in

17 Fayetteville and Camp Lejune in Jacksonville.

18 Personnel from those sites frequently are deployed

19 first to areas of conflict.

20 And I think the civilians in our state,

21 perhaps more than in other areas of the country, are

22 well aware of the necessity to properly protect our

23 troops against the ravages of war. For that reason,

24 we have a particular interest in such things as

25 toxic warfare and biological warfare agents.


1 My own background is pathogenic

2 bacteriology and mycology and the immune responses

3 to those. I taught at North Carolina State

4 University, to veterinary students since 1982 and,

5 prior to that, at the Veterinary College at the

6 University of Illinois and, in the '70's, to medical

7 students at the University of Alabama in Birmingham.

8 My area of expertise is infectious

9 agents. In the -- factors, I've worked with

10 potential BW agents such as Listeria Salmonella,

11 plague bacillus, anthrax, and I've also worked with

12 mycoplasms that aren't known to be BW agents but

13 certainly cause disease in armed forces personnel.

14 I've served on VA committees and

15 committees for the DOD, as you know from my

16 biography.

17 I've been Director of Biotechnology at

18 North Carolina State University, which is an

19 extremely fine program. Ten percent of our faculty

20 in that program that I directed are members of the

21 National Academy of Sciences.

22 The potential for BW agents, using --

23 microorganisms was evident to us in the middle '80's

24 and, for that reason, we've been more involved,

25 perhaps, in those places in studying the potential


1 for that to be used as a weapon against our troops.

2 Because of my knowledge of BW agents,

3 actually beginning in 1965 when we worked with the

4 Navy on Neisseria meningitidis outbreak in

5 Alexandria, Egypt, Commander Gray asked me to serve

6 as an advisor to him.

7 I was interacted with the Veterinary

8 Corps, which is one aspect of my background that I

9 think Commander Gray wanted to take advantage of.

10 I'm familiar, for instance, with the potential cause

11 of piles of animals, which was mentioned in

12 yesterday's testimony, found in the Gulf War site,

13 and also where action was being taken by the

14 Veterinary Corps to study the potential untoward

15 effects of environmental exposure using dogs that

16 were deployed in the Gulf. Those dogs are now,

17 rather than being euthanized, are being maintained

18 for their natural life near San Antonio, Texas.

19 I might just add that canines are an

20 appropriate sentinel animal for potential oncogenous

21 or cancer-causing agents in humans. Unlike cats,

22 tumors in dogs are particular similar to those of

23 humans and have no known bio-etiology.

24 One recommendation I'll make, and I'll

25 make that in writing to your Committee, is, having


1 served on the first committee, which was asked by

2 Secretary of Defense Cheney to be the first group to

3 review a service-wide -- that is, Department of

4 Defense medical RNB command research program, I can

5 say that I would recommend to your committee that

6 greater emphasis be given to the support of

7 appropriate research by the military.

8 And by that I mean, if I can suggest,

9 that the NIH, for instance, is far better equipped

10 to study breast cancer than is the military.

11 However, the military is far better equipped to

12 study vaccine development for potential BW agents

13 and the potential for recombinant bacteria, or

14 fungi, or viruses, for that matter, as potential

15 agents of war in the future.

16 Clearly, as we recognized in the middle

17 '80's, and now I think the general population of the

18 country recognizes, BW agents and chemical agents

19 certainly are the poor countries' atomic bomb and

20 this is an area where I think we've been deficient

21 in pursuing because of concerns in the civilian

22 population of the potential for offensive warfare.


24 much, Dr. Carter.

25 I think what we'll try to do at this


1 point is we'll ask the panel to direct questions to

2 either Dr. Gray and his team or to the panel. And

3 I'm not sure how you want to handle mikes.

4 MS. GWIN: We have a walk-around mike

5 sitting on the corner of the table there.


7 can be passed; and, if it's directed to someone who

8 is not at the table, they can come forward and use

9 the walk-around mike.

10 Andrea?

11 DR. TAYLOR: I have a couple of

12 questions, and I guess I'll start with Dr. Gray,

13 regarding -- well, I understand the limitations are

14 that these are all active duty personnel versus

15 anyone who has left the service, there are no

16 surveys conducted among those?

17 DR. GRAY: Our first three studies

18 involved only active duty personnel. They were

19 exploratory in nature. The next generation of

20 studies include all personnel that were involved in

21 the Gulf, as well as control populations which were

22 not involved but on active duty.

23 DR. TAYLOR: So the information that

24 will be collected, will that be medical information

25 or will it just be self-reported questionnaires on


1 the second group?

2 DR. GRAY: Well, we didn't go into much

3 detail regarding the second studies but Study IV is

4 a survey where we will get some self-reported data

5 there, a mail survey, followed up by telephone

6 interview, following a model which was recommended

7 to us by Dr. Donna Baird, one of our Scientific

8 Advisors, Chicago Health Study. That is an

9 infertility study.

10 We have yet to finalize the second and

11 third phase of that study. We've contemplated

12 actually having a physical exam portion in Phase III

13 should funding continue and we find some interesting

14 things via the telephone interview regarding

15 infertility and pregnancy outcomes.

16 Study V is a survey of the 17,000

17 Seabees and that involves largely a questionnaire

18 and a follow-up with telephone interview of a

19 subsegment. We haven't yet decided if we will look

20 at medical records. The difficulty with medical

21 records review is, of course, they are spread out

22 all over the world in the DOD and again in the non-

23 federal system.

24 Then we have two other studies, Study VI

25 and VII, which involve -- Study VI is looking at


1 California non-federal hospitalization database; and

2 so that's medically oriented -- and data captured

3 for other purposes. And Study VII is a review of

4 the seven actively surveilled birth defect

5 registries.

6 So some of the studies will involve

7 medical records review and some will not, in a

8 nutshell.

9 DR. TAYLOR: I'll come back to that,

10 I'll let somebody ask another question. But I have

11 another question to ask before, it's regarding the

12 serologic tests that were conducted, the blood and

13 urine samples that were collected.

14 You mentioned that tests for

15 leishmaniasis will be done at some point?

16 DR. GRAY: Well, that's an example; and,

17 as yet, there is no good rapid serologic test such

18 as -- for leishmaniasis. Should one become

19 available, we would very much entertain examining

20 our pre- and post-War for serologic evidence of

21 leishmaniasis.

22 DR. TAYLOR: So right as of now, most --

23 those sample have just been collected and they are

24 being stored; right?

25 DR. GRAY: Right, they are stored in


1 five aliquots. And the concept is, as hypotheses

2 arise, we decide through advisors whether or not to

3 expand some of our sera to test those infectious

4 disease relating hypotheses.

5 DR. TAYLOR: One other question

6 regarding the hospitalization data. I'm trying to

7 figure out what would be the purpose of looking at

8 this data and what are you planning to achieve?

9 Because many of the persons who served in the Gulf

10 may not end up in the hospital with chronic fatigue

11 syndrome, that's usually not where that would be

12 reported. They wouldn't be hospitalized for that or

13 for headaches or some of the other.

14 So what are you planning to achieve to

15 get at what the cause of the symptoms, of health

16 symptoms that are currently being felt by Gulf War

17 veterans?

18 DR. GRAY: The concept is largely an

19 exploratory one and a rule-out one. The initial

20 study was to see if there were any differences

21 because we didn't really have good case data. If

22 there was a temporally-related unique exposure that

23 affected a lot of people in the Gulf, we might see

24 increased hospitalizations among the Gulf War

25 veterans. So it was sort of exploratory.


1 Study VI review is -- sort of continues

2 along the same vein. So we're basically looking for

3 hypotheses to test.

4 I think one of the unique things about

5 the available databases are, we could rapidly check

6 certain hypotheses. There was an article, I think

7 in New England Journal regarding aplastic anemias,

8 we were to quickly review that outcome, ICD-9 code,

9 and see if there was an increased risk. And we're

10 looking at other specific categories.

11 Dr. Winkelstein has suggested some

12 approaches for respiratory illnesses and we're

13 pursuing that. And some of the findings that we

14 presented today regarding mental illness and

15 genitourinary outcomes we're examining further.

16 Whether or not we will do case-

17 controlled studies and interviews or bring in people

18 for exams remains yet to be determined.

19 DR. TAYLOR: I still have another

20 question?


22 DR. TAYLOR: And I'm not trying to be

23 sarcastic but, do you have to follow OMB, the same

24 processes the other researchers --

25 DR. GRAY: Yes, we do and we've


1 submitted our package for both studies, the Seabee

2 mail survey and the infertility, to OMB. We think

3 we're in advance -- we received an approval

4 contingent upon making some changes, so we

5 anticipate approval soon.


7 back to the three studies, really, that all have the

8 same problems you presented, the limitation that

9 anyone had to be on active duty. And I just don't

10 know what that tells us about the problem we're

11 trying to explore.

12 If we have sick people, they're probably

13 not on active duty. So I don't know what the

14 conclusions you presented in any of the three

15 studies really mean and what we're to do about that.

16 Could you tell me the justification for

17 doing any studies that are limited to only active

18 duty and also that have a response rate as low as

19 this?

20 DR. GRAY: Well, you must know that we

21 captured, in Studies II and III, the general

22 hospitalization study and the birth rate, we

23 captured all data so, if you're looking at similar

24 participation, we have a hundred percent of those

25 studies.


1 And the concept was --


3 DR. GRAY: Yes.


5 misunderstood. I thought in the hospitalization, it

6 was only in military hospitals.

7 DR. GRAY: That's correct.


9 wouldn't be -- all people who have been separated

10 wouldn't have been hospitalized in military

11 hospitals.

12 DR. GRAY: I didn't mean to infer that.

13 What I meant is, among the people that

14 have remained on active duty, we would capture a

15 hundred percent --


17 question is, that as long as your studies are only

18 of those on active duty, what does it tell us about

19 the general problem we're trying to get at? How

20 does it elucidate the issue of whether we have Gulf

21 War veterans who are ill different than Gulf War --

22 than non-Gulf War?

23 DR. GRAY: I think I agree with what

24 you're inferring, that we cannot extrapolate these

25 data to people who have left the service; we


1 certainly understand that.

2 But, among the populations that remained

3 on active duty, we can examine some potential

4 differences.


6 concern is, of what value is it to know that those

7 on active duty don't have any difference; what does

8 it tell us about the problem we're trying to

9 address?

10 DR. GRAY: I think if we found a marked

11 difference for a causalry it might lead to further

12 illustrative studies.


14 positive finding would have meant something and

15 negative is probably meaningless at this point?

16 DR. GRAY: I wouldn't say it's

17 meaningless but I'd say it would be hard to infer to

18 the population who left the active service.


20 apply to the birth defect data, as well; would it

21 not?

22 DR. COWAN: Yes, it would apply to the

23 birth defect data. I took some comfort in that the

24 rates of attrition were very similar among Gulf War

25 veterans to non-Gulf War veterans, and indeed were a


1 little higher for female Gulf War veterans.

2 However, it may be possible that those

3 people that have a problem, differentially, leave

4 the military and, if they are in small-enough

5 numbers, then we would not be able to detect that

6 with our survival -- our attrition rates.

7 So I agree with you that we cannot

8 extrapolate to them although the evidence wasn't

9 there that they were leaving in disproportionately

10 high numbers.


12 question about the birth defects, at this point, you

13 have not broken those down to look at the

14 differences in the diagnoses of the birth defects

15 between the deployed and non-deployed; is that

16 correct?

17 DR. GRAY: The specific categories?


19 DR. GRAY: Yes, we haven't.


21 You will be doing that and looking to

22 see were there any differences?

23 DR. GRAY: Actually, our Scientific

24 Advisors recommended that that would not be a good

25 use of these data. We have that capability. But


1 they thought that using the actively-surveilled

2 seven-state birth defect registries where they have

3 data beyond just birth, actually out to a year, and

4 good follow-up, good interview data would be much

5 more appropriate.

6 So we are a little hesitant to look at

7 specific ICD-9 codes in those studies.


9 then, that the birth defects data that we have,

10 again, doesn't tell us a great deal but you are

11 planning another study which I do believe, from the

12 protocol I've reviewed, does appear to be one that

13 should give us information. And that's being pilot-

14 tested in Hawaii at the present time; is that

15 correct?

16 DR. ARANETA: We returned from Honolulu

17 eight days ago and the purpose of the pilot study

18 was, first of all, to determine if our ability to

19 match our records of military personnel who could

20 have been deployed or resided or were residents of

21 Hawaii before they joined the military between 1989

22 and through 1993 actually live births in the State

23 of Hawaii.

24 We looked at several matching algorithms

25 to try to maximize identification of the number of


1 births. We are still analyzing the results of that

2 match and the birth defects monitoring program of

3 Honolulu will then identify which of the children

4 born to Persian Gulf War veterans and era veterans

5 were diagnosed with the congenital anomaly.

6 The advantage of Study VII is that,

7 first of all, since there is no national birth

8 defect surveillance system, there are, however,

9 seven states which participate in active

10 surveillance for birth defects -- for major birth

11 defects. And these seven states represent 18

12 percent of all the births in the United States so we

13 will be able to capture one-fifth of all the births

14 in the U.S.

15 We will also be -- and since these birth

16 defects registries ascertain birth defects through

17 the first year of infancy, we will be able to

18 identify births which are missed at birth. Only

19 about 60 to 70 percent of all birth defects are

20 diagnosed at birth. The remaining 30 to 40 percent

21 are diagnosed during -- anywhere between the first

22 month through the 12 months of infancy. So it will

23 enhance case ascertainment.

24 Secondly, since they are population-

25 based, it will enable comparisons of the prevalence


1 and the rates of birth defects among military

2 personnel who are still on active duty, those who

3 had separated, and those in the general non-military

4 population.


6 much; I think that is helpful. And, from the

7 material we received before and the reviews, it does

8 appear that that study will be a meaningful study if

9 the pilot works out and you're able to do what you

10 hope to be able to do. And we'll be interested in

11 following that.
12 And, Dr. Cowan, did you have --

13 DR. COWAN: Yes, Dr. Lashof.

14 I don't think the Committee should be

15 left with the impression that the birth defect study

16 suffers from the same limitations that the other

17 study does. Dr. Donna Baird, who is an Advisor also

18 from North Carolina International Institute of

19 Environmental Health Science, who couldn't be here

20 today, would probably support this -- Dr.

21 Winkelstein can make comments.

22 But I think we should point out that

23 most of the people being evaluated, and I see this

24 in my newspaper every day with Fort Bragg being the

25 largest active army based in the United States, of


1 the continuing concern about birth defects. But the

2 birth defects that one's looking at are in the --

3 predominantly in the wives of military and those men

4 are still in the military.

5 So I don't think the concern you

6 initially voiced applies to the same extent in the

7 birth defect study.

8 Warren, did you have --


10 might -- I mean, it's hard to know and, I think,

11 until we get the other study it's hard to be sure.

12 And the reason I have that concern is that, if the

13 men who have separated have separated because of

14 illness and were exposed more heavily to

15 environmental agents over there, then they may

16 indeed be having more problems with -- and so on.

17 So that's the only reason I question

18 whether we can make any flat-footed statements when

19 it's all based on active duty people.

20 DR. COWAN: People who have problems,

21 and I'm speaking specifically to the birth defect

22 problems, would have an incentive to stay in the

23 military because they would continue to receive

24 high-quality medical care for their families.

25 So I don't think that we can, without


1 further evaluation of it, speculate that people that

2 have problems, speaking only of birth defect

3 problems, are differentially leaving the military.

4 I think that there is a disincentive for them to

5 leave.


7 certainly true once they have a child with a birth

8 defect, but whether they are sick -- husbands have

9 left before they've had the baby, before it has a

10 birth defect, is the issue, that we just don't know.

11 And, until we have more data on troop location and

12 can compare troop location exposure data to look at

13 some of this, it's going to be very difficult to

14 know whether those who remained in versus those who

15 left have had the same experience in the Gulf; that

16 is really my point.

17 But let me now ask you about this

18 additional survey you're planning of 17,000. What

19 is the goal of that study and what will it add to

20 either the Iowa Study or the National Study being

21 carried out of the 15,000 that we heard this morning

22 that is being done by Dr. Kang and his group?

23 DR. DLUGOSZ: The goal of the Seabee

24 health study is to expand and extend upon the

25 original Seabee study and to correct the


1 deficiencies in the original study. The objective

2 is to determine whether Gulf War veterans have a

3 higher rate of symptoms and illness than non-Gulf

4 War veterans.

5 The original study, we've learned a

6 great deal from that. That study's limitation has

7 been well-discussed here, is that it only deals with

8 active duty people, the people that remained on

9 active duty, and does not include Reservists and

10 people who have separated from active duty.

11 Another limitation of that study is,

12 it's cross-sectional in nature and information on

13 exposures is limited to what people have told us.

14 The new Seabee study is a non-concurrent historical

15 prospective study and can take advantage of

16 information from the Registry of Unit Locations,

17 which the Army will provide, I believe next month.

18 And it has a greater size, which can be
19 used -- we are setting up a cohort which can be used

20 to evaluate new hypotheses as they might arise in

21 the future, collect baseline information at this

22 time before we get too far away from the Gulf War.

23 So that answers your question.

24 COMMITTEE CHAIR LASHOF: Well, could you

25 comment further, and then I'd like Dr. Winkelstein


1 and Dr. Morgenstern to comment further, as to what

2 it adds to, and how it relates to the two other

3 studies that are similar that we've heard about

4 yesterday and today, one being the Iowa Study and

5 the one you heard this morning on the National

6 Survey that's being carried out by the VA, which

7 covers 30,000, 15,000 deployed and 15,000 non-

8 deployed.

9 DR. DLUGOSZ: It's somewhat similar in

10 design to the Iowa Study but the Iowa Study, of

11 course, is limited to Iowa. The Seabee health study

12 is limited -- is not limited, it covers personnel

13 all over the world and in the United States. And

14 it's much larger in nature.

15 As compared to the study from the

16 Veterans Administration, the Seabee health study is

17 -- it's a little bit different in that it covers

18 only one cohort, an occupational group of naval

19 construction workers.

20 And the concerns we hear over and over

21 again about Persian Gulf illness, that these

22 illnesses might relate to some environmental

23 exposures or to some things that are not routine

24 parts of battle, as the routine latent effects of

25 war, post-traumatic stress disorder and things.


1 Using the Seabees -- using this entire

2 cohort, I think, is advantageous in that we can look

3 at a group of workers that are highly mobile, that

4 are in many different locations in the Persian Gulf,

5 and are not intensely -- were not intensely exposed

6 to combat and the affects of combat, but they were

7 exposed to the medications, the prophylactic

8 medications, they were exposed to vaccinations, they

9 were exposed to the environmental affects of the

10 Gulf. And that's how I can relate to those other

11 studies.


13 Dr. Winkelstein?

14 DR. WINKELSTEIN: If I understood the

15 protocol, the plan is to do a survey of the entire

16 Seabee roster, if you will, at some point in time,

17 which would include Reserves as well as active duty,

18 and then to do a mail survey, and then, in some way,

19 to follow these people for a period of time.

20 A look in the protocol, it's a little

21 bit unclear to me exactly how long they plan to

22 follow them. In the introduction it talks about the

23 protocol applying only for two years but, if you

24 examine their estimates of sample size and so forth,

25 they are based on the 15-year follow-up.


1 The first major problem that I have with

2 this particular study is the basic survey, which is

3 planned to be a mail survey. And we've heard mail

4 surveys discussed yesterday and today and I think

5 that everyone that's been involved in such surveys

6 realizes that the response rates are not very good.

7 You've got a response rate somewhere in the

8 neighborhood of 50 to 65 percent. Generally

9 speaking, as I've said several times in the last 24

10 hours, that leaves you open for all kinds of

11 problems right there to begin with.

12 Now the protocol, from my point of view,

13 doesn't really deal with the complexities of the

14 follow-up. Now I think the Veterans' study of

15 mortality is reasonably rigorous. I mean, they have

16 the entire population, they have a good way of

17 getting the end points which are death. I mean, you

18 have a national death registry and deaths are

19 generally certified; you can't get buried without a

20 death certificate.

21 So that aspect of a mortality study and

22 mortality follow-up is relatively easy.

23 But I think that the end points that are

24 being proposed to study in the Seabee study are not

25 thank kind of end points. Generally speaking, they


1 are much softer end points. And, also, there's some

2 question as to whether it's appropriate to follow-up

3 on coronary heart disease, which is a major

4 component of the proposal.

5 So I have a lot of problems with the

6 protocol as I've seen it and I would like to be able

7 to study it in more detail and discuss with the

8 investigators at greater length before I made any

9 more comments about it.


11 Dr. Morgenstern?

12 DR. MORGENSTERN: I agree with Dr.

13 Winkelstein's comments about that study.

14 Let me return to your original question

15 to Commander Gray about the value of these other

16 studies that will be done to include inactive former

17 military personnel in the original three that we saw

18 this morning, which is the active military

19 personnel.

20 The response, I believe, was that these

21 are exploratory studies and, therefore, if we see

22 some positive results it's indicative that we have

23 to do more work and pursue the reasons for that;

24 but, if the results are negative, that probably

25 means that there's nothing to it and I guess that


1 means we don't have to do any more work.

2 I disagree with that very strongly and

3 for a whole bunch of reasons that probably would

4 require some elaboration.

5 But one reason is that the problem here

6 is, first, it's not generalizability; that's not the

7 issue of studying active personnel. Can we

8 generalize to inactive personnel; that's not the

9 question. The question is one of bias, how you

10 select the subjects in such a way as to distort some

11 estimate of a causal relationship, presumed causal

12 relationship.

13 And to the extent that we select

14 subjects for the study in ways, on variables that

15 might themselves have been influenced by Gulf War

16 participation, then we introduce a potential bias,

17 particularly if those variables are also related to

18 the outcome of the study.

19 Now it seems to me that that concern is

20 a major one here because it deals with all three of

21 the studies we discussed. I suspect intuitively

22 that it's much more of a problem with the study of

23 symptoms in the Seabees and in the hospitalization

24 than it is in the birth defects. Nevertheless, it's

25 a problem for all.


1 Now I don't think that a positive

2 finding means that we should go further and a

3 negative finding means that we should stop and say

4 there's no effect, it's nothing significant. The

5 fact is that there are many alternative explanations

6 for the negative findings that are consistent with

7 real effects.

8 In any study of occupational or

9 environmental exposures, for example, a major

10 concern, a major source of bias, is

11 misclassification and measurement of key variables,

12 outcome in particular in this case. And, to the

13 extent that there is such measurement error and

14 misclassification, the likely direction of the bias

15 resulting is towards the no, make it appear that

16 there are really no effects when there really could

17 be.

18 There are other concerns, too. So the

19 bias can go in either direction. As a matter of

20 fact, with respect to the hospitalization, it

21 appears that likely bias might have been toward the

22 no because of differences in pre-War health status.

23 The one thing we saw this morning that

24 might indicate that, one of the findings on

25 hospitalization, was how different the deployed and


1 non-deployed were prior to the deployments of the

2 War. There was much more hospitalization in the

3 non-deployed group than in the deployed group. It

4 was a tremendous difference. Did you see how wide

5 apart those curves were?

6 Now, why were they so wide apart? Well,
7 clearly, those who were deployed were healthier in

8 some way. Now we can't ignore that. And, to say

9 that attempting to control for some surrogate proxy

10 confounder didn't correct the problem doesn't mean

11 that we solved the problem, it means that we have to

12 deal with it more thoroughly.

13 To draw conclusions from these studies

14 is going to be very tenuous. And the problem with

15 bias isn't that you know the direction and the

16 magnitude of the bias, the problem is that you don't

17 know. And that means that you have no confidence in

18 drawing conclusions from the results.


20 John?

21 DR. BALDESCHWIELER: If there were a

22 specific -- this is directed at Dr. Gray.

23 If there were specific pathogen or an

24 environmental factor that were unique to service in

25 the Gulf, then a study of active duty personnel, in


1 fact, should have some chance of revealing a

2 difference between the Gulf and non-Gulf groups. In

3 the absence of a positive effect, that still should

4 allow you to place some limits on the occurrence of

5 disease.

6 Can you give us some feeling for the

7 numbers? That is, in your hospitalization study,

8 for example, if you see no positive effect within,

9 say, confidence limits of 90 percent, what are the

10 numbers of occurrences that would be below the limit

11 of statistical significance?

12 DR. GRAY: Well, I think we've listed

13 some of the confidence intervals around the rates.

14 I think your question about the statistical

15 significance requires a couple more elements to

16 figure out exactly what you're at.

17 DR. BALDESCHWIELER: I would like

18 something -- if you could give me just a general

19 feeling for what kinds of limits could you place on

20 the non-occurrence.

21 DR. GRAY: Yeah. If you figure that we

22 had approximately 1.2 million people who were in the

23 cohorts of interest initially, we do have -- I mean,

24 compared to studies that are published every day --

25 a heck of a lot of power for some of the outcomes,


1 particularly those that had a significant frequency.

2 I mean, there's over a thousand ICD-9 codes.

3 So it's a little bit difficult to answer

4 your question --

5 DR. BALDESCHWIELER: Could you, for

6 example, detect the occurrence of a thousand events

7 at a confidence level of 95 percent with the sample

8 size?

9 DR. GRAY: Could we detect the

10 occurrence?

11 DR. BALDESCHWIELER: Yes, the occurrence

12 of some difference between Gulf and non-Gulf service

13 that involved, say, a thousand events.

14 DR. GRAY: So you're saying, if we have

15 2,000 cases in the Gulf War veterans and a thousand

16 in the non-veterans, could we detect that -- could

17 we infer some power? Yes, we can.

18 DR. BALDESCHWIELER: But at a 95 percent

19 confidence level?

20 DR. GRAY: I'm sure we can for that

21 example, that it would detect the difference.

22 DR. BALDESCHWIELER: Could you detect

23 the difference of 500?

24 DR. GRAY: Well, again, it depends on

25 the -- you know, it depends on the number of


1 outcomes in each of the two categories.


3 DR. GRAY: I think -- it's just

4 difficult for me to answer that. We'd be happy to

5 respond to more specific questions at a later date

6 to the Committee. And we could send you power

7 calculations regarding some of the rate differences.

8 But, beyond that, I can't help you really too much

9 here.

10 DR. BALDESCHWIELER: Well, I think first

11 thing, some upper bounds or limits what would be

12 detectable is a potentially useful --

13 DR. GRAY: We would agree with you.

14 I don't think the focus of the

15 hospitalization study is -- say that there was no

16 obvious hospitalization differences between these

17 two groups, hence, let's don't study this any more.

18 Our focus was to see where a quick look

19 at data that were already available, inexpensive,

20 and easy to access to try to answer some of the

21 important questions that our veterans who are

22 concerned about what they showed. And we know that

23 we have -- they are very much limited in what we can

24 say about those data.

25 So I think one of things we'll try to do


1 is qualify that as best we can with respect to the

2 conclusions we do make.

3 DR. BALDESCHWIELER: Well, I think I'm

4 really referring to a different way of presenting

5 the outcomes and, rather than saying that, for

6 example, there's no indication of a positive

7 correlation with service in the Gulf under various

8 categories of diseases, it might be useful to say --

9 I mean, within a 95 percent confidence level, that

10 we can report that there must have been a number

11 fewer than "x", that is less than a certain upper-

12 bound.


14 want to comment?

15 DR. WINKELSTEIN: Well, I'm not a bias

16 statistician so I'm hesitant.

17 But I think one thing that strikes me

18 out of these data are that you have huge numbers and

19 so you can have very, very small relative risks that

20 are statistically significant because of the vast

21 numbers of observations. And that may be

22 meaningless.

23 And, if you have very small amount of

24 bias, the statistical test, of course, doesn't

25 realize that there's bias there so, let's say you


1 had a very tiny bias that produced, let's say,

2 relative risk of 1.2 or 1.3 -- we've seen some of

3 those kinds of relative risks -- those can be

4 statistically significant, meaningless, or biased.

5 So I think we ought to be very cautious

6 in the use of statistical significance with studies

7 like this with such large numbers in the

8 denominators.

9 Maybe Hal is more sophisticated than I

10 am on this subject.


12 DR. MORGENSTERN: I think I've warned

13 the Committee about treating the results of the

14 studies, boiling it down to statistical

15 significance.

16 You hear a lot about this, you see it in

17 the press, you see it in journals, you see it in

18 clinical papers; and it's true, this is a widely-

19 regarded concept. Somehow it's become almost

20 concrete in research findings.

21 Nevertheless, there isn't absolutely a

22 shred of scientific nature to the notion of

23 statistically significant. I truly wish it were

24 struck from the vocabulary of medical/clinical

25 researchers.


1 It is completely arbitrary to say that

2 there is something significant about a finding when

3 the p-value, which is a rather arcane summary

4 statistic, is less than .05 versus when it's greater

5 than .05.

6 And one gets in trouble if one thinks

7 that you can boil down the conclusions of a study by

8 sort of taking this simple statistic and

9 dichotomizing it as significant or not significant.

10 A lot of people do it and it gets done quite a bit

11 and it makes things seem easy but, in fact, it's

12 covering up a lot.

13 For example, this morning I believe we

14 saw -- well, we saw the best estimate of an excess

15 mortality and hospitalization of, what, 13 percent

16 for the Gulf War participants versus the comparison

17 group; am I right? Was that the rate ratio, 1.13

18 overall?

19 MR. COATE: The mortality or the

20 morbidity?


22 hospitalization, overall hospitalization? The rate

23 ratio for the Gulf War participation versus non-

24 participation was a rate ratio 1.13 and confidence

25 interval that just barely crossed the null-value and


1 the p-value is .06?

2 I warn you not to say, not to conclude

3 on the basis of this, that there is no significant

4 relationship. That is a conclusion that, I think,

5 belies the true nature of what statistics are

6 telling us and we have to go further than that.

7 I'm not suggesting that the researchers

8 would not go further than that, nevertheless, that's

9 what going to get reported, very often that's what

10 is going to make the newspapers, and that's what

11 people are going to believe.

12 Now why was there a 13-percent excess

13 hospitalization in Gulf War participants when, in

14 fact, it was much less before the war? Doesn't that

15 strike you as something going on there? Why?

16 And what makes these analyses

17 exploratory is that they often generate more

18 questions than they answer, and that's the question

19 being generated here, and that's the question that

20 we should now pursue: Why was there more

21 hospitalization among Gulf War participants after

22 the Gulf War even though there was considerably less

23 in the same groups before the Gulf War? And,

24 particularly, shortly before the Gulf War.

25 Now I think we have to address that


1 question and, to boil it down to a p-value of less

2 than .05 is entirely misleading.


4 care to ask further questions concerning p-value?

5 DR. BALDESCHWIELER: This is probably

6 not the right forum.

7 But confidence is much better; isn't

8 this true. I mean, it looks like a difference but

9 --

10 DR. MORGENSTERN: Yes, in defense of the

11 research team -- I mean, they did present 95 percent

12 confidence intervals for virtually everything they

13 showed us, most of the things I remember. And the

14 result I even quoted, there was a 95 percent

15 confidence interval so they were basically telling

16 you what the likelihood -- you know, what the band

17 of effects that you would expect, you know, within

18 95 percent probability.

19 Nevertheless, as Dr. Winkelstein pointed

20 out, that assumes no bias. And those confidence

21 intervals basically pretend that there is absolutely

22 no bias and that what you got was perfectly valid.

23 And, of course, that addresses the question that we

24 were talking about before, source of bias.



1 basically, if the sample isn't a good sample, it

2 doesn't matter what statistics you apply to it.

3 DR. MORGENSTERN: Right, right.

4 And, you know, even though this is a

5 huge sample size in many of these studies initially,

6 if one starts weeding it down to subgroups or

7 looking at very rare outcomes like certain kinds of

8 birth defects, you will have a lot less power than

9 you might think.

10 So I agree with you in a sense of there

11 should be power calculations.

12 Now, in our March report we actually

13 called for that and, as I recall, in the response

14 that the research team gave in writing, the written

15 response which occurred a few days later, they

16 agreed with all that and they said they would do it.


18 questions? Any further questions?

19 Dr. Ascher, did you want to make a final

20 comment on this aspect from the point of view of the

21 Epidemiological Board or --

22 DR. ASCHER: I would like to speak to

23 the AFEB issue.

24 I'm a representative of the AFEB; I

25 spoke yesterday. I'll try not to repeat myself.


1 But one of the aspects, observing this

2 overall process, to me is, this resembles a military

3 operation like we did in taking over Grenada, which

4 is, it was a free-for-all. All the forces went in

5 at once, they didn't talk to each other, their

6 radios wouldn't talk to each other, they shot each

7 other; and we didn't really have much of an overall

8 approach.

9 And, in this case, it's a similar thing,

10 that the VA has their studies, the Navy has their

11 studies, the Army has their studies, everybody has

12 their studies but there's no command and control at

13 the top.

14 Now who would be ostensibly responsible

15 for that? Well, that would be a role for the AFEB.

16 Well, up to this point, there has been

17 no official process by which this has occurred but

18 we have had unofficial input. And one of our

19 frustrations, reflected in Dr. Kuller's letter, is,

20 in the previous review of the protocols we've heard

21 today and others, we have made exactly the same

22 comments that were made by the outside advisory

23 panel that was commissioned sitting beside me, by

24 you today, by other people. So we're repeating

25 ourselves.


1 And one of the questions is, unless

2 anybody pulls all of this stuff together into

3 coherent approach of command and control of this,

4 it's going to go on and on.

5 For example, we made a very strong

6 statement that a birth defects monitoring program

7 was not worthwhile at the present time. And we

8 reviewed this and other things that should be done

9 including getting the Reservists in early, getting

10 social factors into the analysis. And apparently

11 there was absolutely no response to those comments.

12 That is our frustration. Dr. Kuller's letter

13 reflects the same thing.

14 In spite of that, they commissioned an

15 outside panel in addition to our review. So, at

16 some point after the free-for-all, let's find out

17 what's good among it, pull it all together, and get

18 it up at the top and make some hard decisions as to

19 what's worth doing.

20 I don't think OMB is where this should

21 meet. OMB is great and they do have a lot of

22 validity in their role, but I don't think they want

23 to get involved in these kinds of things, the big

24 picture.

25 I hope that's the forest, not the trees.



2 much.

3 Are there any closing comments Commander

4 Gray would like to make, or his team, on any of

5 these studies and questions and critiques? It was

6 giving you a tough time.

7 DR. GRAY: We very much appreciate your

8 comments and we try to employ particularly our

9 Scientific Advisory Panel's comments as much as we

10 could. There are certain limitations with respect

11 to questions that we must answer, and finances.

12 But, anyway, we're actually trying to do

13 the very best job we can. We are very concerned

14 because we have, among us -- a number of us are Gulf

15 War veterans and we want to know if there is

16 something out there.

17 We appreciate anything the Committee can

18 do to help clarify things, streamline whatever

19 direction we should go next, and -- well, anyway, we

20 just thank you.


22 much. Any other closing comments?

23 If not, I thank all the panels; I

24 appreciate your input this morning and appreciate

25 your all coming.


1 We, on our panel, the next item on our

2 agenda is to talk about our strategy and where do we

3 go from here.

4 I think, from everything we've heard,

5 it's clear that will be asking staff to follow-up on

6 all the studies we've heard, keep us posted on the

7 progress of those studies. I think we will want to

8 look at some of the smaller studies.

9 I think, in these two days, we have

10 covered the big, major studies; there are a number

11 of others that are in the research plan that bear on

12 epidemiology. There are other studies in the

13 research plan that are more experimental and

14 environmental in the series that we will have to

15 also tackle at some time downstream.

16 I think the next big charge to us to try

17 to put all this together in a way to make some

18 recommendations for our interim report, which is due

19 in February.

20 So my suggestion is that we want to ask

21 staff to follow-up on what we've done today, take a

22 look and prepare for us brief summaries on the

23 additional studies that deal in epidemiology that we

24 didn't get to in the last two days, and that we

25 consider for our December meeting a further


1 discussion of the recommendations concerning the

2 whole series of issues that we've explored over the

3 last several meetings which include the

4 implementation, the outreach issues, and where we

5 stand on the epidemiology studies at this point.

6 I'm open to other suggestions, ideas, or

7 concerns that the subcommittee has.

8 Of course, our next meeting will be of

9 the full Committee and this subcommittee will be

10 reporting to the full Committee on what we came away

11 with.

12 John, do you have any other thoughts or

13 suggestions?

14 DR. BALDESCHWIELER: I think not.


16 not, again I thank you all and we will stand

17 adjourned.

18 (Whereupon, at 11:58 a.m., the meeting was

19 closed.)