992 1 IN THE UNITED STATES DISTRICT COURT 2 DISTRICT OF WYOMING 3 -------------------------------------------------------- 4 THE ESTATES OF DEBORAH MARIE TOBIN and ALYSSA ANN TOBIN, deceased, by 5 TIMOTHY JOHN TOBIN, personal representative; and THE ESTATES OF 6 DONALD JACK SCHELL and RITA CHARLOTTE SCHELL, deceased, 7 by NEVA KAY HARDY, personal representative, 8 Plaintiffs, Case No. 00-CV-0025-BEA 9 vs. May 29, 2001 Volume VI 10 SMITHKLINE BEECHAM PHARMACEUTICALS, 11 Defendant. ----------------------------------------------------------- 12 13 14 TRANSCRIPT OF TRIAL PROCEEDINGS 15 16 Transcript of Trial Proceedings in the above-entitled 17 matter before the Honorable William C. Beaman, Magistrate, 18 and a jury of eight, at Cheyenne, Wyoming, commencing on the 19 21st day of May, 2001. 20 21 22 23 Court Reporter: Ms. Janet Dew-Harris, RPR, FCRR Official Court Reporter 24 2120 Capitol Avenue Room 2228 25 Cheyenne, Wyoming 82001 (307) 635-3884 993 1 A P P E A R A N C E S 2 For the Plaintiffs: MR. JAMES E. FITZGERALD Attorney at Law 3 THE FITZGERALD LAW FIRM 2108 Warren Avenue 4 Cheyenne, Wyoming 82001 5 MR. ANDY VICKERY Attorney at Law 6 VICKERY & WALDNER, LLP 2929 Allen Parkway 7 Suite 2410 Houston, Texas 77019 8 For the Defendant: MR. THOMAS G. GORMAN 9 MS. MISHA E. WESTBY Attorneys at Law 10 HIRST & APPLEGATE, P.C. 1720 Carey Avenue 11 Suite 200 Cheyenne, Wyoming 82001 12 MR. CHARLES F. PREUSS 13 MR. VERN ZVOLEFF Attorneys at Law 14 PREUSS SHANAGHER ZVOLEFF & ZIMMER 225 Bush Street 15 15th Floor San Francisco, California 94104 16 MS. TAMAR P. HALPERN, Ph.D. 17 Attorney at Law PHILLIPS LYTLE HITCHCOCK 18 BLAINE & HUBER, LLP 3400 HSBC Center 19 Buffalo, New York 14203 20 INDEX TO WITNESSES DEFENDANT'S PAGE 21 PHILIP WANG, M.D. Direct - Mr. Preuss 995 22 Cross - Mr. Vickery 1041 Redirect - Mr. Preuss 1136 23 SHERRY MCGRATH 24 Direct - Mr. Gorman 1137 Cross - Mr. Fitzgerald 1175 25 Redirect - Mr. Gorman 1196 Recross - Mr. Fitzgerald 1197 994 1 INDEX TO WITNESSES CONTINUED 2 DEFENDANT'S PAGE JUDITH LAFFERTY 3 Direct - Mr. Gorman 1198 Cross - Mr. Fitzgerald 1212 4 ROBERT HARDY 5 Direct - Mr. Gorman 1220 Cross - Mr. Fitzgerald 1228 6 INDEX TO EXHIBITS 7 PLAINTIFFS' RECEIVED 8 61 1185 9 DEFENDANT'S JJ 1044 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 995 09:06:49 1 P R O C E E D I N G S 09:06:49 2 (Trial proceedings reconvened 09:06:49 3 9:00 a.m., May 29, 2001.) 09:06:49 4 THE COURT: Good morning. I trust everyone had a 09:06:49 5 pleasant holiday weekend. 09:06:49 6 What's your schedule with your witness? 09:06:49 7 MR. VICKERY: She's here, Your Honor. Counsel will 09:06:49 8 go ahead with their case and we'll work her in whenever is 09:06:49 9 convenient with them. 09:06:49 10 THE COURT: Very well. 09:06:49 11 If you would, please, Mr. Preuss, call your next 09:06:49 12 witness. 09:06:49 13 MR. PREUSS: Thank you, Your Honor. Defendant will 09:06:49 14 call Dr. Philip Wang at this time. 09:07:27 15 THE CLERK: Please state your name and spell it for 09:07:27 16 the record. 09:07:27 17 THE WITNESS: My name is Philip Wang, 09:07:27 18 P H I L I P, W A N G. 09:07:27 19 Can you hear me? 20 21 PHILIP WANG, M.D., 22 called as a witness on behalf of the Defendant, being first 23 duly sworn, testified as follows: 24 DIRECT EXAMINATION 09:07:37 25 Q. (BY MR. PREUSS) Good morning, Mr. Wang. 996 09:07:41 1 A. Good morning. 09:07:42 2 Q. Are you a physician, sir? 09:07:43 3 A. Yes, I am a physician. 09:07:44 4 Q. And where do you practice, sir? 09:07:46 5 A. I practice -- well, I don't practice now but I am from 09:07:49 6 Boston, Massachusetts. 09:07:50 7 Q. And where do you work, sir? 09:07:51 8 A. I work at the Brigham and Women's Hospital at Harvard 09:07:57 9 Medical School. I have an appointment at Harvard Medical 09:08:00 10 School in psychiatry, in medicine and health care policy. 09:08:04 11 I'm also an instructor of epidemiology at the Harvard School 09:08:09 12 of Public Health. 09:08:09 13 Q. Could you outline for us your educational background, sir, 09:08:13 14 beginning with college. 09:08:15 15 A. Sure. I attended Harvard College and graduated in 1984 09:08:18 16 with a degree in biochemistry. I received my medical degree 09:08:21 17 from Harvard Medical School in 1989. I completed my 09:08:25 18 psychiatry residency in -- at the Beth Israel Hospital which 09:08:30 19 is a Harvard teaching hospital in 1993. 09:08:33 20 After that I attended graduate school in epidemiology 09:08:37 21 and received my Master's degree from the Harvard School of 09:08:41 22 Public Health in 1996. 09:08:42 23 And I received my doctoral degree in epidemiology, 09:08:46 24 again from the Harvard School of Public Health in 1998. 09:08:51 25 Q. Are you board certified, sir? 997 09:08:53 1 A. Yes, I'm board certified in psychiatry. 09:08:56 2 Q. And what does it mean to be board certified, sir? 09:08:59 3 A. Board certification means you've passed several exams set 09:09:02 4 up by the board that's meant to certify psychiatrists. One 09:09:07 5 is a written exam. Another is an actual exam with patients 09:09:11 6 to make sure that you are competent in the care of 09:09:14 7 psychiatric patients. 09:09:16 8 And I passed my board certification in 1994. 09:09:19 9 Q. You indicated that you had a residency in psychiatry at 09:09:23 10 Beth Israel Hospital, sir. How long was that residency? 09:09:27 11 A. The residency is for four years. It also includes a 09:09:29 12 one-year medical internship as well. 09:09:32 13 Q. And was that residency in general psychiatry? 09:09:37 14 A. It was in general psychiatry. 09:09:39 15 Q. And I think you indicated that after you got your 09:09:42 16 residency program completed you then embarked upon further 09:09:46 17 education in the area of epidemiology; is that right? 09:09:49 18 A. Yes. After I completed my training in psychiatry, I then 09:09:52 19 went to graduate school in epidemiology for five years. 09:09:58 20 Q. And that culminated in your doctorate in epidemiology? 09:10:01 21 A. Yes. I received both a Master's degree and also a 09:10:04 22 doctorate in epidemiology. 09:10:07 23 Q. You indicated you were an instructor at Harvard Medical 09:10:10 24 School, sir? 09:10:11 25 A. Yes. 998 09:10:12 1 Q. And who do you instruct or teach, sir? 09:10:15 2 A. Well, I teach at two levels. I teach in classrooms. I 09:10:20 3 teach the introductory psychiatric epidemiology course at the 09:10:25 4 Harvard School of Public Health. 09:10:26 5 I also teach at the individual level. I instruct 09:10:29 6 physicians who are in clinical fellowships on how to do 09:10:33 7 psychiatric -- pharmacoepidemiologic research. I currently 09:10:38 8 have five physicians that I'm teaching. Three are 09:10:43 9 neurologists -- three are oncologists, one is a neurologist 09:10:47 10 and one is a general medical physician. 09:10:49 11 Q. And oncology involves cancer? 09:10:52 12 A. Sorry. Oncologists are doctors who specialize in 09:10:58 13 treating cancer. Neurologists are specialists who treat 09:11:02 14 patients with nerve disorders. And a general medical doctor 09:11:05 15 is probably the doctor you would see for all of the other 09:11:09 16 sort of nonspecialized conditions. 09:11:14 17 Q. Do you hold any hospital appointments, sir? 09:11:17 18 A. Yes, I'm an associate physician at the Brigham and Women's 09:11:20 19 Hospital and I'm an associate psychiatrist at the Beth Israel 09:11:25 20 Hospital in Boston. Both of these are teaching hospitals of 09:11:27 21 Harvard Medical School. 09:11:29 22 Q. When you say teaching hospitals, what do you mean, sir? 09:11:32 23 A. They're centers where Harvard Medical School actually 09:11:36 24 trains its medical students, so they come to these hospitals 09:11:39 25 for practical training in medicine. 999 09:11:44 1 Q. And do you do research work, sir? 09:11:47 2 A. Yes, I do. 09:11:47 3 Q. And what type of research do you do? 09:11:54 4 A. I do three types of research. The first type of research 09:11:54 5 I do is called pharmacoepidemiology. And it is a long word. 09:11:58 6 You will hear us say it several times. It is the study of 09:12:01 7 whether drugs cause adverse reactions. And I specifically 09:12:05 8 study psychiatric medication. That's one type. 09:12:08 9 The second type of research I do is I look at how -- 09:12:13 10 it is called health services research. I look at how 09:12:16 11 psychiatric medications are used and I try to understand who 09:12:20 12 is getting these medications, are they getting appropriate 09:12:23 13 treatment, are these drugs being underused, overused, 09:12:27 14 misused. 09:12:29 15 There's a third type which is called randomized 09:12:32 16 controlled trials designed to see if we can improve how 09:12:36 17 psychiatric medications are used. 09:12:38 18 Q. And who pays for your research work, sir? 09:12:41 19 A. My research is almost exclusively funded by the National 09:12:46 20 Institute of Mental Health which is a federal agency, and 09:12:49 21 also I receive funding from the McArthur Foundation. 09:12:53 22 Q. And what is the National Institute of Mental Health, sir? 09:12:57 23 A. The National Institute of Mental Health, again, is part of 09:12:59 24 the federal government and it is responsible for doing two 09:13:03 25 things, really. One is to funnel research dollars into 1000 09:13:12 1 projects, to try to have research projects conducted in areas 09:13:15 2 that are in need of good work. 09:13:18 3 They also try to set the agenda, so if the National 09:13:21 4 Institutes of Health thinks that a certain area may not have 09:13:25 5 enough information in it, they may actually try to funnel 09:13:30 6 researchers into that area by, again, providing funds to do 09:13:34 7 research. 09:13:35 8 Q. Have you received any funding from SmithKline or Glaxo? 09:13:39 9 A. No, never. 09:13:40 10 Q. Have you published articles in the scientific journals? 09:13:43 11 A. Yes, I have. 09:13:44 12 Q. About how many, sir? 09:13:45 13 A. Approximately now about 40 publications in the 09:13:50 14 peer-reviewed literature. 09:13:51 15 Q. And what do you mean by peer-reviewed literature? 09:13:56 16 A. By peer review what I mean is journals when you submit an 09:14:00 17 article to them for publication will have three of your 09:14:03 18 peers, three other scientists, maybe two, review your work. 09:14:07 19 And you don't know who they are, so you're blind to who the 09:14:10 20 reviewers are. 09:14:12 21 They will make suggestions to you about how to 09:14:14 22 improve your work. They will try to catch errors. And they 09:14:18 23 may reject your paper if they think it is really awful. 09:14:21 24 So it is in place -- peer review is in place in order 09:14:25 25 to keep the standards high so what ends up being published in 1001 09:14:31 1 the literature is correct and high quality. 09:14:33 2 Q. And what journals? Can you give us a smattering of 09:14:37 3 journals you've published your articles in? 09:14:39 4 A. Yeah, I can do that. I've published in the Journal of the 09:14:43 5 American Medical Association, JAMA; the American Journal of 09:14:48 6 Psychiatry, the Journal of Clinical Epidemiology, the Journal 09:14:51 7 of General Internal Medicine. A variety of journals. 09:14:56 8 Q. And are you yourself a reviewer for certain of these 09:14:59 9 journals, sir? 09:15:00 10 A. Yes, I routinely review. I'm one of the peer reviewers 09:15:04 11 for people who submit their articles to journals. 09:15:08 12 Q. Are you a member of any professional organizations, sir? 09:15:11 13 A. I'm a member of the International Society for 09:15:13 14 Pharmacoepidemiology. I'm a member of the American 09:15:18 15 Psychiatric Association. I'm a member of the Massachusetts 09:15:20 16 Psychiatric Society. 09:15:23 17 Q. Do you serve on any national committees for the American 09:15:28 18 Psychiatric Association, sir? 09:15:29 19 A. Yes, I've served on three national committees that were 09:15:32 20 assembled by the American Psychiatric Association to create 09:15:37 21 practice guidelines. What these documents are are they're a 09:15:43 22 review of the entire scientific literature and they're 09:15:47 23 intended to present what the best, most effective and safe 09:15:51 24 treatment of psychiatric patients should be. 09:15:54 25 And the three committees that I serve on, I was a 1002 09:15:58 1 consultant to the work group that wrote the practice 09:16:02 2 guideline for the treatment of patients with depression. I 09:16:05 3 was also a consultant on the work -- to the work group that 09:16:10 4 prepared the practice guideline for the treatment of patients 09:16:14 5 with delirium, another condition. 09:16:16 6 I was also a consultant to the work group that put 09:16:19 7 together the practice guideline for the treatment of patients 09:16:21 8 with schizophrenia, another type of mental disorder. 09:16:24 9 Q. Are these guidelines then distributed to clinicians, 09:16:28 10 psychiatrists and others that may treat patients with 09:16:31 11 psychiatric disorders? 09:16:33 12 A. Yes, that's the intended purpose of these guidelines. 09:16:39 13 They're meant for the frontline practicing psychiatrist who 09:16:39 14 is seeing patients. And, again, it is to present to them 09:16:43 15 what the -- what is safe and effective treatment. And these 09:16:51 16 guidelines have been widely disseminated in both the United 09:16:55 17 States and actually internationally as well. 09:16:57 18 Q. Do you serve on any committees for the National Institutes 09:16:59 19 of Mental Health, sir? 09:17:01 20 A. Yes, actually next month I will serve on a special review 09:17:05 21 panel established by the National Institute of Mental Health 09:17:10 22 to review grants that it receives. 09:17:14 23 Q. Are they grants or are they grant proposals? 09:17:17 24 A. Proposals for grants. I shortened it to grants. The NIMH 09:17:22 25 receives lots of grant proposals, and they have limited 1003 09:17:25 1 funds. They can't give everybody who applies for a grant 09:17:29 2 funding. 09:17:31 3 So the NIMH, the National Institute of Mental Health, 09:17:36 4 establishes special review manuals to determine how to best 09:17:40 5 use the money. The money is all of ours. It is tax dollars. 09:17:44 6 What they use the special review panels for is to determine 09:17:47 7 which of the proposals for grants really deserve the money. 09:17:56 8 Q. I understand you recently received an award from the 09:17:59 9 American Psychiatric Association called the Health Services 09:18:01 10 Research Award. What was that for, sir? 09:18:03 11 A. It was specifically for my research on psychiatric 09:18:07 12 medications. 09:18:10 13 Q. Now, have you ever testified or worked on any other legal 09:18:15 14 case such as the one we're here today on? 09:18:18 15 A. No, no, I never have. 09:18:19 16 Q. This is the first time? 09:18:20 17 A. Yeah, it is the first time. It is actually quite hard to 09:18:24 18 do, to be away from my work, number one, and also my family, 09:18:30 19 my wife, and I just had a child, so it is tough to sort of 09:18:34 20 leave home. 09:18:38 21 Q. Doctor, you've used the word "epidemiology" and we've 09:18:43 22 heard about that in the last few days as the trial has 09:18:46 23 progressed. 09:18:46 24 Can you tell us what epidemiology is, sir? 09:18:50 25 A. Sure. Epidemiology is a scientific discipline in which 1004 09:18:57 1 the scientist, the epidemiologist is trying to see whether 09:19:04 2 the exposure -- by exposure I mean things like drugs, foods, 09:19:09 3 toxins in the environment -- whether these exposures cause 09:19:15 4 problems, problems like disease or adverse events. This is 09:19:19 5 what the field tries to study. 09:19:22 6 Q. Can you give us a example that would be an everyday 09:19:26 7 example of an epidemiology study? 09:19:28 8 A. Yeah. Maybe some of you or most of you are aware that 09:19:31 9 from earlier epidemiologic studies butter was shown -- if you 09:19:37 10 eat too much butter it was shown to be related to developing 09:19:40 11 heart disease and things like heart attacks. 09:19:43 12 That was older research. For all of us here who have 09:19:47 13 switched to margarine, there's now newer studies which show 09:19:51 14 that margarine may not be any better than butter. So these 09:19:56 15 are the types of studies that epidemiologists do. 09:20:00 16 Q. All right. How about pharmacoepidemiologists? What do 09:20:04 17 they study, sir? 09:20:05 18 A. It is a tough word. I have trouble saying it too. 09:20:08 19 Pharmacoepidemiologists specialize in studying prescription 09:20:12 20 medications. They try to -- they study whether prescription 09:20:18 21 medication use is associated with or causes problems, 09:20:23 22 disease, adverse conditions. 09:20:24 23 Q. And that's what you are, right? 09:20:26 24 A. Yes, sir, that is what I do. 09:20:30 25 Q. And what are they looking for? 1005 09:20:32 1 A. There's really two things they're trying to do. There's 09:20:34 2 two tasks. One task is pharmacoepidemiologists are trying to 09:20:41 3 identify whether drugs really do cause problems, whether they 09:20:45 4 really do pose public health threats. That's one thing they 09:20:50 5 try to identify. 09:20:51 6 They have an equally important task which is 09:20:55 7 identifying wrong claims about drugs, wrong claims that drugs 09:21:00 8 cause adverse events and these are two equally important 09:21:04 9 tasks. 09:21:05 10 Q. Why is the latter task, looking for wrong claims, as you 09:21:09 11 referred to it, important in terms of public health? 09:21:14 12 A. Well, if -- wrong claims can cause damage, especially with 09:21:19 13 psychiatric medications. Psychiatric medication use is 09:21:23 14 stigmatized and so there's tremendous underuse by people who 09:21:27 15 need treatment. 09:21:30 16 And to give you an example of this, I published a 09:21:33 17 study last year in the Journal of General Internal Medicine. 09:21:36 18 I was the first author. We examined psychiatric medication 09:21:39 19 use in the entire U.S., and we looked specifically at 09:21:51 20 patients with depression and patients with anxiety disorders. 09:21:51 21 And what we observed was among these types of 09:21:54 22 patients with these disorders, depression and anxiety 09:21:56 23 disorders, only about 14 percent were receiving treatment 09:22:00 24 that you could in any shape or form call minimally adequate. 09:22:05 25 14 percent. 1006 09:22:06 1 We have also subsequently followed up and found that 09:22:09 2 among the people who do get treatment -- again, very few get 09:22:14 3 treatment, but among people who do get treatment, there are 09:22:17 4 tremendous delays. People really hesitate and wait before 09:22:20 5 they try to seek care. 09:22:22 6 We found on average people wait 11 years before 09:22:26 7 having their first symptom and actually seeking some kind of 09:22:31 8 treatment. And so this is -- you can imagine, then, what the 09:22:36 9 damage would be if you had wrong claims about a drug. People 09:22:41 10 who are -- there's already underuse and if there's wrong 09:22:45 11 claims out there, there may be even more underuse by people 09:22:50 12 who really need the treatments. 09:22:51 13 If there's wrong claims about a drug causing an 09:22:54 14 adverse event, there may be longer delays, even longer than 09:22:59 15 11 years. Even among people who do get treatment, they may 09:23:04 16 needlessly worry if their drug causes an adverse event. 09:23:09 17 Q. Doctor, could you tell us how epidemiologists go about 09:23:13 18 figuring out whether or not a drug causes a specific side 09:23:17 19 effect or disease process? 09:23:19 20 A. Sure. It is a long-established, well-established method. 09:23:25 21 And I have a little diagram here. Is it okay if I draw on 09:23:28 22 the -- 09:23:29 23 Q. If you would, please. Maybe turn it around so that 09:23:31 24 everybody can see. 09:23:37 25 MR. PREUSS: Your Honor, we previously marked that as 1007 09:23:40 1 Defendant's Exhibit JJ. 09:23:43 2 THE WITNESS: Can everyone see? Can everyone hear 09:23:48 3 me? Please tell me if you can't hear me. 09:23:52 4 Q. (BY MR. PREUSS) Well, let's start at the beginning, then, 09:23:55 5 Doctor. What's the first thing to think about? 09:23:58 6 A. I'll try to speak up so you can hear me. What I'm writing 09:24:02 7 here first is questions and I'm writing here on the 09:24:07 8 right-hand side answers. And this ugly thing I'm drawing 09:24:18 9 here -- you will have to excuse my artistic abilities. This 09:24:23 10 is supposed to be a river. So that's a river. 09:24:28 11 Q. You have a river then between the questions and the 09:24:30 12 answers? 09:24:32 13 A. Questions and answers. And let me explain why I put this 09:24:35 14 thing up here. 09:24:36 15 Pharmacoepidemiologists especially start by asking 09:24:40 16 questions. They start by asking questions about medications: 09:24:44 17 Is it possible that a medication causes an adverse event? Is 09:24:48 18 it possible that a medication doesn't cause the adverse 09:24:50 19 event? These are the types of questions 09:24:52 20 pharmacoepidemiologists start with. 09:24:56 21 They try to answer those questions. They try to 09:24:59 22 actually show, actually show whether the medication causes 09:25:03 23 the adverse event or whether the drug actually doesn't cause 09:25:07 24 the adverse event. They actually seek answers. 09:25:11 25 I put the river in here because this is the task of 1008 09:25:17 1 the epidemiologist. The task of the epidemiologist is to go 09:25:20 2 from questions concerning whether drugs cause adverse events 09:25:24 3 over the river to getting their answers: Yes, the drug 09:25:28 4 really does cause the adverse event; or no, the drug actually 09:25:33 5 doesn't cause the adverse event. So hopefully that's clear. 09:25:38 6 Where do epidemiologists get their questions from? 09:25:42 7 There's a couple of places and I'll put them down. I'm first 09:25:47 8 writing animal data -- I meant animal studies. 09:25:58 9 The second thing I'm writing here is case reports. 09:26:02 10 And I'll define both of these. 09:26:04 11 Animal studies are studies where a scientist will 09:26:09 12 take mice or rats, give them a drug and see whether there's a 09:26:13 13 certain effect. 09:26:16 14 Epidemiologists look at these animal studies and ask 09:26:19 15 questions. The questions they ask are well, if there was an 09:26:24 16 effect in rats or mice, might there also be an effect in 09:26:28 17 humans? These are the kinds of questions an epidemiologist 09:26:31 18 will get from animal data. 09:26:34 19 They also use case reports to generate their 09:26:37 20 questions. What's a case report? A case report is a 09:26:40 21 publication usually from a physician who is practicing and 09:26:45 22 the physician who has, you know, a practice, patients, will 09:26:48 23 say, "You know, I saw one of my patients who was on a 09:26:52 24 medication. They also happened to develop a condition or an 09:26:56 25 illness." And they will wonder, they will say, "I wonder if 1009 09:27:02 1 the drug caused the illness or the condition or the adverse 09:27:06 2 events?" 09:27:08 3 They raise the question and so they write up this 09:27:17 4 case report and get it published in order to stimulate people 09:27:17 5 like epidemiologists to also ask the question and do the 09:27:20 6 studies. This is how epidemiologists get their questions. 09:27:26 7 How do epidemiologists go from the questions from 09:27:29 8 animal studies, questions from case reports over the river to 09:27:33 9 actually answering the question of whether the drug does or 09:27:37 10 actually doesn't cause the adverse event? 09:27:40 11 Well, there's one preferred, best way to do it and 09:27:45 12 I'll write it down here. What I'm writing is randomized 09:28:02 13 controlled trials. 09:28:03 14 Q. Can you explain both randomized and controlled? 09:28:06 15 A. Sure. I will spend some time explaining both what 09:28:10 16 randomized is, what controlled is and why they're so 09:28:13 17 important. 09:28:14 18 Randomized means the epidemiologist is actually doing 09:28:19 19 an experiment. They take their subjects and they randomly 09:28:24 20 give some subjects drug A. They will then take -- randomly 09:28:33 21 give other subjects things like placebo or sugar pill -- I 09:28:37 22 don't know if you've used the term "placebo." 09:28:40 23 Q. We have. 09:28:42 24 A. Sugar pills. They take another group and give randomly 09:28:45 25 drug B, a second drug. Why do they randomly assign who is 1010 09:28:49 1 getting drug A, placebo, drug B? They do it because they 09:28:54 2 don't want to load the dice against one of the drugs, in 09:28:58 3 favor or against one of the drugs. They want to make sure 09:29:01 4 the same type of people are getting drug A, drug B and sugar 09:29:05 5 pill. They really want to make sure there's no bias here. 09:29:09 6 That's why they use randomization. The scientist is 09:29:13 7 randomly saying who gets drug A, drug B and sugar pill. 09:29:17 8 MR. VICKERY: Excuse me, Doctor. Your Honor, I know 09:29:19 9 that Dr. Wang has not testified before, but I would ask the 09:29:23 10 Court to instruct him he's not here to ask questions like, 09:29:27 11 "Is that clear?" but to merely testify in response to 09:29:30 12 questions by Mr. Preuss. 09:29:32 13 MR. PREUSS: That's fine. 09:29:33 14 THE COURT: That's fine. Go ahead. 09:29:36 15 THE WITNESS: Sorry. 09:29:37 16 A. What does controlled mean? Sorry. 09:29:41 17 THE WITNESS: Is this the type of question you mean 09:29:43 18 don't ask? 09:29:44 19 THE COURT: It is all right. Go ahead. 09:29:46 20 Q. (BY MR. PREUSS) If you could define for us what 09:29:48 21 controlled means, that would be great. 09:29:52 22 A. Controlled means within your study there's a group of 09:29:55 23 patients who are not receiving the drug you're interested in 09:29:58 24 studying. So there's a group that's not getting the drug 09:30:02 25 you're most interested in observing whether it causes or 1011 09:30:04 1 doesn't cause an adverse event. The reason why this control 09:30:09 2 group is so critical is because of background rates of 09:30:13 3 illness, background rates of conditions. 09:30:16 4 Background rates means there could be a -- a 09:30:21 5 condition can occur just by chance. It can occur whether or 09:30:26 6 not you're on a drug. In the case of suicide, there is a 09:30:29 7 background rate. Suicide has been occurring since time 09:30:35 8 immemorial. Suicide has been occurring long before 09:30:38 9 antidepressants were ever produced, marketed. 09:30:42 10 And to give you an idea of how common this background 09:30:45 11 rate is, if you look at all of the deaths in the United 09:30:48 12 States, one out of 75 deaths that occur is by suicide. 09:30:52 13 That's what I mean by background rate. It occurs whether or 09:30:55 14 not you're on antidepressants. It can occur by chance or due 09:30:58 15 to depression or other conditions. 09:31:01 16 The background rate gets even higher amongst patients 09:31:06 17 with depression. If you look at the number of people who die 09:31:10 18 with -- among depressed populations, one out of seven deaths 09:31:15 19 is by suicide, so it is very common. There's a high 09:31:18 20 background rate. 09:31:19 21 In order for epidemiologists to know whether a drug 09:31:22 22 causes a rate of adverse event or a rate of disease higher 09:31:27 23 than the background rate, they need some way to figure out 09:31:31 24 what the background rate is. What is the rate of having 09:31:34 25 suicide without the drug? 1012 09:31:36 1 And that's why a control group is so critical. If 09:31:39 2 you don't have a control group you don't know what the 09:31:42 3 background rate is of developing suicide or any other adverse 09:31:45 4 condition just by chance or due to depression or something 09:31:48 5 else. 09:31:49 6 And by having a control group, you can then say does 09:31:53 7 the drug cause a little bit more suicide or a little bit less 09:31:57 8 suicide or the same amount of suicide as in the control 09:32:00 9 group, as in the baseline, the background rate? 09:32:05 10 This, again, is the preferred means. This is what we 09:32:10 11 call the gold standard, the best way to help the 09:32:13 12 epidemiologist to cross from the questions to actually 09:32:16 13 getting the answer does a drug really or really not cause an 09:32:25 14 adverse event. 09:32:25 15 There's a second type of study. It is not the best, 09:32:27 16 it is secondary. It is -- and I'll tell you what I'm 09:32:31 17 writing -- observational study, and putting another line at 09:32:49 18 the bottom, if adjusted -- if adjusted for bias. 09:32:56 19 Q. For what? 09:32:57 20 A. For bias. And I'll explain all of these terms. 09:33:03 21 Q. Could you start with the observational study and define 09:33:05 22 that for us, please. 09:33:07 23 A. Sure. Observational means it is not a randomized study. 09:33:11 24 It is not an experiment. The investigator is not randomly 09:33:15 25 assigning who gets drug A, drug B or the sugar pill. The 1013 09:33:21 1 investigator is not ensuring that the same type of person is 09:33:25 2 getting drug A, drug B or the placebo. 09:33:28 3 Instead, in an observational study the investigator 09:33:32 4 simply looks at what is happening in a clinic somewhere or 09:33:38 5 maybe in a hospital or amongst doctors out in private 09:33:42 6 practice. And in the real world there can be -- if there's 09:33:47 7 not randomization, if you're just looking at how doctors 09:33:52 8 prescribe in a local clinic or your local hospital, they 09:33:55 9 prefer to give certain drugs to certain people. 09:33:58 10 And in the case of medications like Paxil, Paxil is 09:34:02 11 given to patients who are at higher risk of suicide for the 09:34:06 12 following reason: It is safer in overdosage. You can take a 09:34:10 13 lot of it and not kill yourself if you overdose on it. 09:34:15 14 Other antidepressants, if you take a lot of them, 09:34:18 15 they will stop your heart and you'll die. So out in the real 09:34:21 16 world in practices, in clinics and hospitals, doctors will 09:34:28 17 sort of assess how high the risk is for the patient to 09:34:31 18 overdose, try to commit suicide. 09:34:34 19 If they think the patient is someone who is really 09:34:37 20 going to go out and overdose, which drug do you think they'll 09:34:41 21 select -- sorry, that was a question. They will try to 09:34:44 22 select the medication that the patient will be safe on in 09:34:47 23 overdose. 09:34:48 24 So if you do an observational study, you will have 09:34:51 25 this bias, this prescribing bias where the patients who are 1014 09:34:57 1 getting drugs like Paxil will be the highest risk patients, 09:35:03 2 the ones who are -- the doctor thinks is most likely to 09:35:06 3 get -- to commit suicide. 09:35:09 4 So these aren't randomized studies. They're 09:35:12 5 observational. What the epidemiologist does is they have to 09:35:15 6 adjust for that prescribing bias. If they don't adjust for 09:35:19 7 the prescribing bias, it is not possible to get an answer. 09:35:28 8 So let me just put another thing over here, 09:35:32 9 observational study, and what I'm writing down here is not 09:35:45 10 adjusted for bias. I said not adjusted for bias. And this 09:35:59 11 means if someone does an observational study -- 09:36:02 12 Q. Excuse me, Dr. Wang. 09:36:06 13 MR. VICKERY: Could we please proceed on questions 09:36:08 14 and answers instead of a lecture from the doctor? 09:36:10 15 THE COURT: Well, he is a professor and I see a lot 09:36:13 16 of that. We've had a lot of that throughout this trial, but 09:36:16 17 I would appreciate questions and answers instead of the 09:36:18 18 dialogue. 09:36:20 19 Q. (BY MR. PREUSS) Doctor, you indicated there on the left 09:36:22 20 that on the question side you have observational study, not 09:36:25 21 adjusted for bias. Can you tell us what you mean by that? 09:36:30 22 A. Sure. I've explained what observational studies are. 09:36:35 23 They're studies where the epidemiologist isn't randomly 09:36:38 24 assigning who is getting the drug or the sugar pill. They're 09:36:42 25 watching what happens out in the real world. 1015 09:36:45 1 And it could have this bias prescribing where certain 09:36:50 2 drugs are prescribed to patients by doctors who are at higher 09:36:54 3 risk. If there's no adjustment for the bias prescribing, 09:36:58 4 these observational studies aren't useless. They can help 09:37:02 5 you ask questions. You can wonder whether drugs cause 09:37:04 6 adverse events or you can wonder if they don't, but these 09:37:08 7 types of studies cannot help you like studies that are 09:37:12 8 adjusted for the bias prescribing in terms of answering these 09:37:16 9 questions, you know, crossing the river from question to 09:37:19 10 answer. 09:37:27 11 Q. Can you give us an example where you recently worked on a 09:37:31 12 study where a question was asked and an answer was obtained? 09:37:35 13 A. Sure. I may have to stand back up there because I can't 09:37:38 14 see the board myself. 09:37:40 15 Q. Sure. 09:37:41 16 A. We recently did a study, it is going to be published in 09:37:46 17 this month's -- it is not June -- the June issue of the 09:37:50 18 Journal of Clinical Epidemiology, and it is a study of 09:37:54 19 whether antidepressants cause breast cancer in women. 09:38:00 20 This study -- the question, do antidepressants cause 09:38:03 21 breast cancer in women -- this question came first from 09:38:08 22 animal studies. There was a scientist, Brandis, up in 09:38:13 23 Canada, who found if he gave rats and mice antidepressants 09:38:17 24 like tricyclics, fluoxetine -- not Paxil, the other 09:38:23 25 antidepressants -- he found that the rats and mice developed 1016 09:38:27 1 breast lesions, the female rats and mice. 09:38:31 2 That generated the question does this happen in 09:38:33 3 humans. After this was published, the animal studies, some 09:38:38 4 case reports occurred which doctors out in practice said, 09:38:41 5 "You know, I had a patient on antidepressants and they also 09:38:45 6 got breast cancer. I wonder if antidepressants really cause 09:38:50 7 breast cancer." The case reports also generated the 09:38:53 8 questions. These are where we got our questions from. 09:38:56 9 So we proceeded to do -- we tried to do this: We 09:39:00 10 tried to say is there -- is there a randomized controlled 09:39:05 11 clinical trial that has been done or one that we could do in 09:39:09 12 order to study this question? This would have been our 09:39:11 13 preferred means, the best way to answer the question. 09:39:14 14 We decided it would have taken too long to wait. We 09:39:19 15 would have had to give people -- you know, women 09:39:22 16 antidepressants now, randomized some women not to 09:39:26 17 antidressants and wait because it takes many, many years 09:39:31 18 before the cancers occur. And this is too important a 09:39:34 19 question to wait that long. If antidepressants do cause 09:39:37 20 breast cancer, you need to find out fast, soon. 09:39:40 21 There was no existing randomized controlled clinical 09:39:44 22 trial for us to use. We would have preferred to use data 09:39:47 23 already collected from an existing study but there wasn't any 09:39:51 24 available. So we went to this type of study, this 09:39:53 25 observational study. 1017 09:39:55 1 But what is crucial is we had to adjust -- if I can 09:39:58 2 back up for a second, in an observational study you're not 09:40:01 3 randomly saying who is going to get an antidepressant, who is 09:40:06 4 not. In an observational study you watch how doctors in 09:40:10 5 practice are giving out antidepressants. 09:40:12 6 But women who are at higher risk of developing breast 09:40:16 7 cancer, believe it or not, are also more likely to get 09:40:20 8 antidepressants from their doctors for a whole bunch of 09:40:25 9 reasons. 09:40:25 10 We had to adjust for the prescribing bias in the real 09:40:28 11 world. We had to take that into account. When we did, the 09:40:31 12 results showed there was no association. Antidepressants do 09:40:35 13 not cause breast cancer. That's an example of the type of 09:40:38 14 studies I do. 09:40:40 15 Q. To review a couple of points, can animal studies by 09:40:43 16 themselves get you to an answer? 09:40:44 17 A. No. Anything -- again, I drew that river there and I hope 09:40:50 18 it is useful to show that anything on this side of the river, 09:40:55 19 on your right-hand side of the river can generate questions 09:41:01 20 about whether drugs cause adverse events, but they can't 09:41:05 21 actually show it. They can't answer the question for you. 09:41:09 22 They can't prove whether the drug really does or doesn't 09:41:13 23 cause the adverse event. 09:41:14 24 Q. And what about adverse experience reports that a company 09:41:17 25 gets with respect to experience that a physician or a patient 1018 09:41:23 1 themself may report by way of adverse experience to the 09:41:26 2 company? Is that a question? 09:41:28 3 A. Adverse event reports are like case reports. Doctors in 09:41:34 4 practice notice, again, that maybe one of their patients was 09:41:37 5 on a drug and also seemed to develop a condition. So you can 09:41:42 6 think of them as case reports. 09:41:44 7 But they're not even as good as case reports. Case 09:41:46 8 reports are published in the scientific literature, often 09:41:49 9 peer reviewed. And so they're held to a higher standard. 09:41:53 10 An adverse event reported, you can literally call up 09:41:57 11 the FDA -- and actually I haven't done that. I assume you 09:42:01 12 can call up the FDA or the drug company and simply say, "I 09:42:04 13 saw a patient yesterday who was taking drug X and they 09:42:07 14 developed, you know, condition Y. Do you guys record this, 09:42:12 15 keep track of it?" There's not as much rigor or information 09:42:17 16 as in a published case report where you can actually read it 09:42:21 17 in a scientific journal. 09:42:24 18 Q. Doctor, we're here today to discuss the issue of whether 09:42:27 19 or not Paxil can cause suicide. Are you aware of any studies 09:42:32 20 on the answer side of the type that you put on the answer 09:42:35 21 side that answer the question as to whether Paxil causes 09:42:39 22 suicide? 09:42:40 23 A. Answer to the question or -- could you repeat that? 09:42:43 24 Q. Sure. Are there any questions on the answer side that 09:42:48 25 would confirm that Paxil causes suicide? 1019 09:42:55 1 A. I'm sorry. You said questions and answer. 09:42:55 2 Q. I'm sorry. 09:42:55 3 A. I'm not sure I'm following your question. 09:42:57 4 Q. Are there any studies that you're aware of that would be 09:43:01 5 on the answered side of your diagram that establish that 09:43:07 6 Paxil causes suicide? 09:43:09 7 A. No, there are no studies on your left-hand side that show 09:43:12 8 that Paxil causes suicide. 09:43:17 9 Q. Maybe my right-hand side -- your left-hand side? 09:43:20 10 A. You're right. Why don't we use my left-hand side and my 09:43:25 11 right-hand side. This side, my left-hand side of the river, 09:43:30 12 there are no studies on that side that show that Paxil causes 09:43:34 13 suicide. 09:43:36 14 Q. All right. And let me ask you the converse of that. Are 09:43:38 15 there any studies that would appear on the answer side that 09:43:43 16 establish that Paxil does not cause suicide? 09:43:46 17 A. Yes, there are. There are several, and if I could, I 09:43:52 18 would like to just list them. 09:43:53 19 Q. Would you please list those and then we'll talk about each 09:43:56 20 one of them. 09:44:09 21 A. I will use a different color and I'll list the names of 09:44:13 22 the studies so when we go through them we can all see what 09:44:17 23 I'm referring to. 09:44:18 24 Lopez-Ibor, a single name just hyphenated; 09:44:24 25 Montgomery, Kahn, Verkes, Paxil healthy volunteers, Sheehan 1020 09:44:55 1 and Dunbar. 09:44:57 2 Q. If we could, first, Doctor, could you turn your attention 09:45:00 3 and tell us about the Lopez-Ibor study and how that answers 09:45:05 4 the question that Paxil does not cause suicide? 09:45:10 5 A. Sure. Lopez-Ibor is a metaanalysis of randomized 09:45:17 6 controlled clinical trials. 09:45:19 7 Q. Let me stop you right there. Tell us what a metaanalysis 09:45:22 8 is, if you would, please. 09:45:24 9 A. A metaanalysis is a term that simply means a combination, 09:45:28 10 a collection of studies. We talked about randomized 09:45:34 11 controlled clinical trials as being the preferred means for 09:45:36 12 answering questions about whether drugs cause adverse events. 09:45:41 13 It is sometimes an advantage to, instead of looking 09:45:44 14 at a single randomized controlled clinical trial, to then 09:45:49 15 take the data from multiple ones -- instead of just one which 09:45:53 16 might be too small, you collect several, multiple randomized 09:45:58 17 controlled clinical trials and you use all of them, if 09:46:01 18 possible. That's what a metaanalysis means. It means taking 09:46:04 19 a collection of randomized controlled clinical trials. 09:46:07 20 Lopez-Ibor did that. They used all the randomized 09:46:11 21 controlled clinical trials in the worldwide database of Paxil 09:46:15 22 trials that were conducted prior to the approval of Paxil. 09:46:20 23 And this involved thousands of patients. The reason why they 09:46:24 24 did the metaanalysis was to get lots of experience from a lot 09:46:28 25 of different patients. 1021 09:46:30 1 They were interested in a few outcomes. I will 09:46:33 2 describe one. One outcome was reductions in suicidality. 09:46:41 3 They were wondering if during a trial suicidality went down. 09:46:46 4 They compared people who were given Paxil to people given 09:46:51 5 placebo, the sugar pill. And Paxil was significantly 09:46:55 6 consistently better than the sugar pill at reducing 09:46:59 7 suicidality. 09:47:01 8 Q. Let me ask you a couple questions there if I might, 09:47:04 9 Doctor. You used the term "suicidality." What does that 09:47:09 10 mean? 09:47:10 11 A. By that I mean they had a rating scale in there, a way to 09:47:12 12 measure how suicidal someone was and the person was reporting 09:47:16 13 about their suicidal ideas. 09:47:18 14 Q. So suicidality refers to suicidal thoughts or ideas? 09:47:23 15 A. In this study and in this specific analysis I'm talking 09:47:27 16 about they were measuring suicidal thoughts. 09:47:31 17 Q. And then you used the term "statistically significant." 09:47:35 18 What does that mean in epidemiologic terms, sir? 09:47:39 19 A. Let me give you a mix of epidemiologic and try to also 09:47:42 20 give you terms that are not epidemiologic because it is a 09:47:47 21 tough concept. 09:47:48 22 In any study you will -- that compares, say, one drug 09:47:53 23 to another drug, you will get a difference. Maybe drug A -- 09:48:00 24 people given drug A have a higher rating than the patients 09:48:04 25 given drug B. 1022 09:48:06 1 Or maybe the patients given drug A make more 09:48:11 2 behaviors of some sort than the subjects given drug B. You 09:48:17 3 will see a difference on the rating score or number of 09:48:20 4 behaviors. 09:48:20 5 The epidemiologists ask -- they have to ask this 09:48:24 6 question, is this a real difference? Is this real or is it 09:48:27 7 just due to chance or error? That's what statistical 09:48:36 8 significance helps you determine. When an epidemiologist 09:48:39 9 says they've calculated that this difference is statistically 09:48:42 10 significant, what they're saying is drug A really is higher, 09:48:49 11 really is higher than drug B on the rating scale or it really 09:48:50 12 is higher on the number of behaviors or whatever they're 09:48:53 13 measuring. 09:48:54 14 Let's say the epidemiologist does the calculation and 09:48:58 15 says this is not a statistically significant difference, it 09:49:01 16 is not statistically significant. What they're telling you 09:49:06 17 there is no difference. What they're observing could be due 09:49:09 18 to chance or error. That's a definition. 09:49:12 19 Q. All right. And now you indicated that they were looking 09:49:14 20 at reduced suicidal thoughts as one of the examination points 09:49:20 21 on the Lopez-Ibor study? 09:49:23 22 A. That was one of the analyses they did. What they found, 09:49:27 23 to repeat it, they found that Paxil was statistically 09:49:30 24 significantly better at reducing the suicidal thoughts than 09:49:34 25 people given the sugar pill. 1023 09:49:36 1 Q. Were there other areas that were looked at in the study? 09:49:40 2 A. Yeah, they did another very interesting and important 09:49:43 3 analysis. They looked not only at the entire -- all the 09:49:47 4 people involved in all of the studies, but they also looked 09:49:50 5 at specifically a subsection of -- a slice of the patients 09:49:54 6 who participated in all of these trials, just a slice of 09:49:59 7 them. The slice they were interested in looking at were 09:50:02 8 patients who at the beginning of the trial had no suicidality 09:50:06 9 at all, no suicidal thoughts. 09:50:09 10 And what they were interested in was studying whether 09:50:12 11 among people free of suicidality at the beginning of the 09:50:18 12 trial -- they don't have any -- does new suicidality emerge, 09:50:23 13 does it come out during the trial. 09:50:25 14 To do that they study it again. To reiterate, people 09:50:30 15 free of the suicidality, this subsection, this slice, what 09:50:35 16 they found was Paxil was statistically significantly better 09:50:38 17 at preventing the emergence of new suicidality, suicidal 09:50:46 18 thoughts, in patients who were free of it at the beginning of 09:50:49 19 the trial than people receiving the sugar pill. 09:50:52 20 Q. Does that cover the Lopez-Ibor study? 09:50:55 21 A. Yes. 09:50:56 22 Q. The next one is the Montgomery study. Can you tell us 09:50:58 23 what the examination points were on that study and what the 09:51:03 24 answers were to that? 09:51:05 25 A. The Montgomery study was similar. They used the clinical 1024 09:51:09 1 worldwide database of randomized controlled trials. They had 09:51:14 2 similar interests so they looked at how well suicidal 09:51:17 3 thoughts were reduced in the entire study population. 09:51:21 4 And they, again, found that Paxil was statistically 09:51:27 5 significantly better than placebo at reducing suicidal 09:51:31 6 thoughts among everybody. And they found using three 09:51:34 7 different measures -- in this study they had three different 09:51:38 8 ways of measuring suicidal thoughts. 09:51:40 9 They did a similar study to the Lopez-Ibor where they 09:51:44 10 also looked at just the slice that was free of suicidal 09:51:48 11 thoughts at the beginning of the trial. None of these people 09:51:52 12 were suicidal. They did not have suicidal thoughts and they 09:51:55 13 just watched to see how much suicidal thought came out during 09:51:59 14 the trial. 09:51:59 15 And they again found that the -- the people given 09:52:04 16 Paxil had statistically significantly less new suicidal 09:52:09 17 thoughts emerge during the trial compared to the people given 09:52:12 18 sugar pill. 09:52:13 19 They also had an interesting result that on one of 09:52:16 20 their scales, one of their measures of suicidal thoughts, 09:52:20 21 they actually found that Paxil did better than another 09:52:24 22 antidepressant, a tricyclic antidepressant. 09:52:30 23 Q. Does that cover then the Montgomery study? 09:52:33 24 A. Yes. 09:52:33 25 Q. And could you move to the Kahn study, please, sir? 1025 09:52:37 1 A. The Kahn study is also a randomized controlled clinical 09:52:43 2 trial metaanalysis. It is up in my left-hand top corner. It 09:52:47 3 is the preferred type of study. It is a collection of these 09:52:52 4 randomized controlled clinical trials. 09:52:56 5 And this one was using all of the randomized 09:52:59 6 controlled clinical trials that the FDA has on file for new 09:53:02 7 antidepressants, again involving thousands of patients. 09:53:06 8 The results of that study are interesting. They 09:53:09 9 had -- they were looking at a slightly different outcome. 09:53:13 10 They looked at actually suicide, actually killing yourself, 09:53:19 11 and they also looked at making suicide attempts, not just 09:53:23 12 having suicide thoughts but actually acting, making suicide 09:53:26 13 attempts, actually killing yourself. 09:53:29 14 What they found was numerically, just if you look at 09:53:33 15 the numbers, Paxil users, people randomized to Paxil made 09:53:38 16 fewer attempts than people given placebo or other 09:53:46 17 antidepressants. 09:53:46 18 They also -- numerically just in terms of numbers, 09:53:50 19 fewer of them actually committed suicide. 09:53:54 20 And let me just say, these results did not achieve 09:53:57 21 statistical significance. So all that means is there's -- 09:54:01 22 you can't say with any certainty that the Paxil -- that Paxil 09:54:05 23 was better at preventing suicides or suicide attempts, but it 09:54:09 24 certainly was no worse. 09:54:13 25 Q. Then could you move to the Verkes study, please? 1026 09:54:16 1 A. Sure. The Verkes study is a single randomized controlled 09:54:21 2 clinical trial, and it is a study in a high-risk population. 09:54:27 3 By high risk what I mean is these patients were all patients 09:54:31 4 who had made a prior suicide attempt, and patients who make 09:54:36 5 suicide attempts are very much at risk for making a 09:54:40 6 subsequent one. They're one of the populations that are 09:54:43 7 maybe at highest risk of making suicide attempts. 09:54:46 8 And in this trial, it was a year-long trial, they 09:54:51 9 randomly assigned people to either Paxil or placebo, the 09:54:55 10 sugar pill, and they found -- they were interested in three 09:54:59 11 different types of outcomes. One of the outcomes they were 09:55:03 12 interested in was the proportion of patients that made a 09:55:07 13 subsequent attempt. Everyone has made an attempt in the past 09:55:10 14 recently. What they were interested in was how many people 09:55:13 15 went out and made another attempt during the year-long trial. 09:55:17 16 What they found was Paxil users were -- they made an 09:55:19 17 adjustment, they had to adjust for one factor called the -- 09:55:24 18 they had to adjust for the previous number of suicide 09:55:28 19 attempts. When they did that, they found that Paxil users 09:55:32 20 were statistically significantly less likely to make a 09:55:35 21 subsequent suicide attempt compared to the people given sugar 09:55:39 22 pill. That's one outcome. 09:55:41 23 The second outcome they were looking at was speed. 09:55:44 24 They wanted you to know were people faster or slower at 09:55:47 25 making their suicide attempts on Paxil versus sugar pill. So 1027 09:55:53 1 it is speed of making the suicide attempt. 09:55:56 2 What they found when they did this analysis, again, 09:55:59 3 Paxil users were slower to make a subsequent suicide attempt 09:56:03 4 than people getting sugar pill. 09:56:06 5 A third interesting study that they did, analysis in 09:56:10 6 this randomized controlled clinical trial was they measured 09:56:14 7 anger, anger scores. They put in a measure of how much anger 09:56:20 8 someone had, a scale called the anger expression inventory, 09:56:24 9 and they observed that people given Paxil after two weeks 09:56:30 10 actually had lower anger scores. So very quickly patients 09:56:34 11 had lower anger scores if they were given Paxil than the 09:56:38 12 patients given placebo, sugar pill. 09:56:44 13 Q. And that covers the Verkes study? 09:56:46 14 A. That covers it. 09:56:47 15 Q. And then you have the Paxil healthy volunteer study. What 09:56:50 16 did your analysis of that study indicate, sir? 09:56:54 17 A. The Paxil healthy volunteer study refers to a group of 09:57:00 18 studies called healthy volunteer studies. When a 09:57:05 19 pharmaceutical company is applying for a new drug approval 09:57:10 20 from the FDA -- FDA is the governmental drug regulatory body. 09:57:16 21 When they're applying for a new drug approval, they 09:57:21 22 need to submit certain types of studies for the FDA to 09:57:24 23 review. One of them are called early phase clinical trials. 09:57:29 24 And in these early phase clinical trials the drug is randomly 09:57:34 25 given -- again these are randomized controlled clinical 1028 09:57:38 1 trials -- the drug is given to healthy volunteers. 09:57:42 2 Healthy volunteers are also -- some of them are 09:57:44 3 randomly given the sugar pill, placebo. What they're really 09:57:48 4 interested in -- they're not interested in these studies in 09:57:51 5 how well the drug works in terms of treating depression or 09:57:54 6 anything like that, because these are all healthy people, 09:57:57 7 healthy volunteers. What they're interested in is how does 09:58:02 8 the drug -- excuse me -- does the drug cause adverse events. 09:58:07 9 And this is a useful database for the following 09:58:09 10 reason: In all of the clinical trials involving over a 09:58:13 11 thousand patients that were taking Paxil, they found not a 09:58:20 12 single case of patients developing suicidal thoughts. They 09:58:25 13 found not a single case among patients given Paxil, healthy 09:58:29 14 volunteers -- they found not a single case of someone making 09:58:32 15 a suicide attempt. Not a single case of someone given Paxil 09:58:36 16 who actually completed suicide. Not a single case of Paxil 09:58:40 17 patients having homicidal thoughts. Not a single Paxil 09:58:50 18 patient made homicide gestures or actually tried to kill 09:58:52 19 someone and no one treated it. 09:58:54 20 Q. Doctor, before we move on to the Sheehan and Dunbar 09:58:58 21 studies, with respect to the studies we've just reviewed, is 09:59:02 22 there anything in those studies that would suggest that Paxil 09:59:05 23 causes suicidality? 09:59:06 24 A. No. As I review it there's nothing to suggest that Paxil 09:59:10 25 causes suicide. In fact, what I've been reviewing for you 1029 09:59:14 1 suggests that Paxil is a good treatment for preventing newly 09:59:19 2 emergent suicide and also an effective treatment for reducing 09:59:23 3 suicidality. 09:59:24 4 Q. Doctor, we've heard some testimony that suggests that 09:59:28 5 Paxil may cause an increase in anxiety. Are there any 09:59:32 6 studies that looked at that issue? 09:59:37 7 A. Those are the two studies I put at the bottom of the list. 09:59:40 8 Again, they're randomized controlled clinical trials, the 09:59:44 9 preferred best way to try to answer the question, get over 09:59:46 10 the river to answering whether drugs cause adverse events. 09:59:51 11 Sheehan is, again, a metaanalysis of randomized 09:59:56 12 controlled clinical trials, and they, again, were examining 09:59:59 13 the worldwide clinical trial database of Paxil studies, again 10:00:04 14 involved with thousands of patients. 10:00:05 15 And they were interested in several different 10:00:09 16 outcomes. One outcome was somatic anxiety. Another outcome 10:00:15 17 was psychic anxiety. A third outcome was agitation, and I 10:00:21 18 will explain what each of those means. 10:00:23 19 Q. Would you please do somatic anxiety and then psychotic? 10:00:28 20 A. Okay. Somatic anxiety means it is a type of anxiety 10:00:36 21 symptom that shows itself or manifests itself through some 10:00:40 22 effect on your body, like if you -- let's say you have this 10:00:43 23 anxiety symptom. You might get sweaty palms or you might 10:00:49 24 feel your stomach be upset, or you might get short of breath 10:00:53 25 or have rapid heartbeat. 1030 10:00:56 1 These are all manifestations of being anxious, but 10:00:59 2 they manifest themselves through -- think of it as your body. 10:01:05 3 Psychic anxiety is a subjective sense of feeling 10:01:10 4 anxious, you feel anxious or scared. That's what psychic 10:01:14 5 anxiety means. 10:01:15 6 Agitation, have you covered agitation? 10:01:18 7 Q. Go ahead and define it as you understand it. 10:01:20 8 A. It is basically self-explanatory. You're agitated. It is 10:01:24 9 a little more straightforward. These are the three outcomes 10:01:27 10 you're interested in. 10:01:31 11 Among patients randomly given Paxil, they found 10:01:36 12 compared to patients given placebo that Paxil users had 10:01:41 13 statistically significant less somatic anxiety, statistically 10:01:46 14 significant less psychic anxiety, and statistically 10:01:51 15 significant less agitation when you compared Paxil users to 10:01:57 16 patients given the placebo pill; so actual reductions in the 10:02:01 17 anxiety and agitation ratings. 10:02:12 18 They also did analysis where they looked at whether 10:02:15 19 Paxil was better than other antidepressants, other 10:02:18 20 tricyclics. They found on one of their measures, agitation 10:02:21 21 at four weeks, at least, there was a significant -- Paxil was 10:02:25 22 significantly better than the tricyclic, the other type of 10:02:29 23 antidepressant. 10:02:30 24 Q. That covers Sheehan, then? 10:02:33 25 A. I don't want to get into too much, but they also had one 1031 10:02:36 1 of these newly emergent analysis where they looked not at the 10:02:41 2 entire population but just the slice of people free of 10:02:44 3 agitation at the beginning of the trial, because here what 10:02:47 4 they wanted to see was did new agitation emerge among people 10:02:53 5 who never had it at the start of the trial. 10:02:56 6 When they compared people randomized to Paxil to 10:03:00 7 people randomized to placebo, they found less newly emergent 10:03:05 8 agitation in this small slice, subsection of people who 10:03:09 9 didn't have agitation at the beginning of the trial. 10:03:13 10 Q. Does that complete Sheehan? 10:03:15 11 A. Yes. 10:03:16 12 Q. Tell us about Dunbar. 10:03:17 13 A. It is a similar study to Sheehan using similar data. It 10:03:20 14 is a metaanalysis of randomized controlled clinical trials, 10:03:24 15 and Dunbar looked at adverse event reporting during the 10:03:31 16 clinical trial. 10:03:31 17 And during the clinical trial when you come in, you 10:03:35 18 know, for your visit, you know, weekly or biweekly or 10:03:39 19 whatever during the trial, the observer will rate whether you 10:03:45 20 report that you've had an adverse event. And the types of 10:03:50 21 adverse events they were reporting were the subjects saying 10:03:53 22 they felt anxiety, that they felt agitation, that they felt 10:04:01 23 nervous. 10:04:01 24 And there was also something called central nervous 10:04:04 25 system stimulation, but it is another type of adverse event, 1032 10:04:08 1 and it basically -- all of these are used to mark whether 10:04:13 2 patients are developing these anxiety-type symptoms. 10:04:19 3 And what they found is when they compared Paxil users 10:04:24 4 to people given sugar pill to people given this other 10:04:30 5 antidepressant tricyclic, there was no statistically 10:04:35 6 significant difference amongst any of the treatments. There 10:04:38 7 was no difference between getting Paxil and getting a sugar 10:04:41 8 pill in terms of reporting any of these adverse events during 10:04:45 9 the trial. 10:04:46 10 Q. All right. Do either Sheehan or Dunbar, then, sir, 10:04:51 11 establish that Paxil causes people to be more agitated? 10:04:55 12 A. No, it doesn't establish that. In fact, as I just 10:05:01 13 described, it actually appears from these studies, these 10:05:04 14 studies support that Paxil is a good treatment for anxiety 10:05:10 15 symptoms, a good treatment for reducing anxiety symptoms and 10:05:14 16 agitation, and it also may be effective treatment for 10:05:17 17 preventing newly emergent anxiety symptoms -- agitation 10:05:22 18 symptoms. 10:05:23 19 Q. There's one study we haven't discussed and that's the 10:05:28 20 Donovan study. Are you familiar with that? 10:05:31 21 A. Yes, I am. 10:05:32 22 Q. Where would that fit on your chart, sir? 10:05:34 23 A. Let me also write it in. I have written Donovan in under 10:05:47 24 observational studies, not adjusted for bias down in my 10:05:51 25 right-hand left-hand corner. 1033 10:05:52 1 Q. Can you tell us what kind of a study it was and why you 10:05:55 2 have put it on the question side? 10:05:58 3 A. Sure. This is -- an example might be good here. When I 10:06:03 4 was flying from Denver to Cheyenne, I -- we hit some 10:06:07 5 turbulence and the airplane -- it was quite actually a 10:06:13 6 troubling ride. 10:06:14 7 And during that time I was thinking there are two 10:06:17 8 things this plane better have to stay up. One is this plane 10:06:20 9 better be well equipped; and two, there better be a pilot who 10:06:26 10 knows how to handle the turbulence, who is aware of the 10:06:30 11 potential dangers and pitfalls and has experience dealing 10:06:33 12 with them. 10:06:34 13 Epidemiologic studies are exactly the same. In order 10:06:37 14 for them to fly, you need two things. You need the study to 10:06:42 15 be properly equipped. All the equipment has to be there. 10:06:45 16 Second thing you need is you need an investigator, or 10:06:49 17 a pilot, if you will, who is skilled and experienced, knows 10:06:54 18 what the dangers are, knows what the pitfalls are. 10:06:57 19 I raise that example because Donovan, unfortunately, 10:07:03 20 is not properly equipped. That's why I put it under that not 10:07:06 21 adjusted for bias. 10:07:08 22 But fortunately, Donovan is an experienced pilot, 10:07:13 23 experienced investigator who knows the pitfalls, the danger 10:07:18 24 in the study and alerts the reader to them. 10:07:22 25 If -- actually, if you bear with my example, Donovan 1034 10:07:26 1 basically kept the plane on the ground and tells you it is 10:07:30 2 not going to fly. 10:07:34 3 Let me tell you why I'm saying all of this. I 10:07:37 4 explained what an observational study is. There's no 10:07:40 5 randomized study. There's no making sure the people 10:07:43 6 randomized to the drugs are the same, the randomized 10:07:48 7 exposures. It is an observation of how doctors prescribe in 10:07:51 8 a single clinic in Chester, England. And remember, drugs 10:07:57 9 like Paxil because they're safer in overdosage, Doctors like 10:08:01 10 to give them to the patients who they think are most at risk 10:08:07 11 of overdosing and committing suicide. 10:08:10 12 They found proof of this prescribing bias. There's 10:08:13 13 actually statistical evidence saying, "We actually observed 10:08:16 14 this bias prescribing." What they observed was in their data 10:08:20 15 Paxil and other SSRIs were being given to patients that were 10:08:26 16 sicker, having difficulty in treatment, switching around a 10:08:30 17 lot on antidepressants. 10:08:32 18 They also found that these SSRIs, Paxil, these types 10:08:36 19 of drugs were being given to patients who had prior histories 10:08:40 20 of suicide attempts. Again, these are high-risk people. And 10:08:45 21 these were given more than other antidepressants. So drugs 10:08:48 22 like Paxil were preferentially biasedly given to patients at 10:08:54 23 higher risk of suicide. 10:08:55 24 They weren't able to adjust for this bias. They 10:08:58 25 couldn't fix it. That's what I mean by underequipped. But 1035 10:09:05 1 fortunately when I said Donovan was a skilled, experienced 10:09:14 2 investigator who knew that this problem was occurring, they 10:09:14 3 warned the reader explicitly that this problem is there, and 10:09:17 4 they also warn the reader, keep this plane on the ground. 10:09:20 5 This study cannot be used to help you cross the river over 10:09:24 6 into getting an answer about whether antidepressants cause 10:09:29 7 suicide or not. They explicitly tell you you can't use this 10:09:33 8 study to establish that. 10:09:34 9 Q. Doctor, is this the quote that you had in mind on that? 10:09:38 10 A. Yeah. Don't take my word for it. Let me read to you 10:09:41 11 their warning. This is specifically what they say. And I'm 10:09:45 12 referring to page 556 of their article. 10:09:50 13 Here's their warning about the bias prescribing: 10:09:53 14 "Prescribers are heeding advice to prescribe 10:09:56 15 safer-in-overdose antidepressants to patients who are 10:10:00 16 perceived to be at greater risk of deliberate self-harm. 10:10:04 17 This effectively loads the dice against antidepressants such 10:10:09 18 as the SSRIs so that this manifests as an apparent excess of 10:10:13 19 self-harm behavior in patients who have been prescribed these 10:10:16 20 antidepressants." 10:10:17 21 It is a pretty clear explanation. He actually does a 10:10:21 22 much better job of explaining the prescribing bias to people 10:10:24 23 at higher risk of overdosage than I think I did. But this is 10:10:28 24 a clear explanation of the problem, the pitfall of the study. 10:10:33 25 But even more importantly, Donovan tells you this 1036 10:10:38 1 study should not be used to establish whether antidepressants 10:10:42 2 cause suicide. It is biased. There was no adjustment, no 10:10:49 3 fixing of the prescribing bias. 10:10:52 4 I'm referring to a quote on page 555. Here is where 10:10:55 5 they explicitly tell you you can't use this to cross the 10:10:59 6 river. "It is difficult to attribute the cause of deliberate 10:11:04 7 self-harm behavior to antidepressant treatment when such 10:11:08 8 behavior also occurred spontaneously during the course of 10:11:11 9 progressive illnesses. Establishment of cause and effect for 10:11:13 10 the different apparent risk of deliberate self-harm 10:11:15 11 associated with different antidepressants seen in the study 10:11:18 12 is, therefore, almost impossible." 10:11:21 13 What Donovan is saying is because this study wasn't 10:11:24 14 equipped -- he smartly is saying keep this plane on the 10:11:27 15 ground, it can be used to generate questions about 10:11:31 16 antidepressants and suicide, but it can't be used to 10:11:34 17 establish whether antidepressants cause suicide. That's why 10:11:38 18 I didn't put it down in that right-hand bottom corner. 10:11:46 19 Q. Your right-hand bottom corner? 10:11:48 20 A. Right, my right-hand bottom corner, down there. 10:11:51 21 Q. Does Donovan then lend any support to the claim that Paxil 10:11:55 22 causes suicide, sir? 10:11:56 23 A. No, it does not, because, again, as the authors instruct 10:12:02 24 you, this study cannot be used to establish that 10:12:06 25 antidepressants cause suicide. They explicitly tell you it 1037 10:12:10 1 can't be done. 10:12:15 2 Q. Now, Doctor, we've been talking about Paxil studies so 10:12:21 3 far, correct? 10:12:23 4 A. That's correct. 10:12:24 5 Q. Now, if you're going to study Paxil and it is whether or 10:12:28 6 not it can cause suicide, can you just borrow studies that 10:12:31 7 deal with Prozac or other SSRIs? 10:12:34 8 A. No, you can't. That's actually very -- it would be a big 10:12:38 9 problem to do that. Drugs in the same class might share some 10:12:42 10 of the same beneficial effects. Drugs in the same 10:12:46 11 antidepressant class might share the beneficial effect of 10:12:50 12 treating depression, depression symptoms, but they don't 10:12:52 13 share the same side effects. 10:12:55 14 And if you assume that, it can get you into a lot of 10:12:58 15 trouble. And I will give you an example of that. There's a 10:13:02 16 medication used to treat psychotic symptoms. It is an 10:13:06 17 antipsychotic. 10:13:08 18 Psychotic symptoms are symptoms like having 10:13:12 19 hallucinations, being paranoid, and they come up when someone 10:13:16 20 has a disease, usually like schizophrenia. There's a 10:13:21 21 particular antidepressant -- antipsychotic medication called 10:13:26 22 clozapine and it is a special drug, really a very good one 10:13:30 23 for treating these symptoms, but it has a rare side effect 10:13:33 24 that's devastating. It causes you to lose your white blood 10:13:38 25 cells which help you fight infection. So in a rare person 1038 10:13:43 1 given clozapine, this antipsychotic medication, they can 10:13:49 2 develop overwhelming infection. 10:13:51 3 Other atypical antipsychotic medications, other drugs 10:13:59 4 in this class, don't share that side effect. What is the 10:14:03 5 danger in assuming that all drugs in the same class share the 10:14:06 6 same side effect? The dangers are as follows: If you assume 10:14:11 7 that clozapine had the same side effects as the other 10:14:15 8 antipsychotic medications in the same class, the danger would 10:14:21 9 be you would miss all of these cases of lowered white blood 10:14:28 10 cell counts and these horrible infections that would occur 10:14:28 11 because you haven't realized that clozapine may have a 10:14:30 12 special different side effect and you've just assumed it has 10:14:33 13 the same side effects as all the others. 10:14:36 14 At the same time, there's also a danger in assuming 10:14:40 15 the other antipsychotic medications act like clozapine. That 10:14:45 16 danger is probably equally as problematic because what would 10:14:48 17 happen if you assumed these other drugs which don't cause 10:14:51 18 this lowering of your white blood cell count? What if you 10:14:55 19 assumed they did? You might needlessly worry the patients 10:15:00 20 taking these antipsychotic medications. 10:15:03 21 As I told you earlier, particularly antipsychotic 10:15:06 22 medications are extremely underused. The people that would 10:15:10 23 benefit from them and need them don't take them. 10:15:13 24 If you say there's a side effect to them that doesn't 10:15:15 25 really exist, you're going to get even fewer people to take 1039 10:15:19 1 the antipsychotic medication who would benefit from it and 10:15:24 2 really need it. 10:15:25 3 Q. We've heard testimony regarding case reports involving 10:15:28 4 Prozac by Teicher and Rothschild. Are you familiar with 10:15:32 5 those? 10:15:32 6 A. I'm familiar with them. 10:15:33 7 Q. Do they shed any light on whether Paxil might cause 10:15:37 8 suicide, sir? 10:15:39 9 A. No, they don't. And for actually at least two reasons. 10:15:43 10 One is they don't involve Paxil. They involve Prozac. We're 10:15:49 11 talking about a different drug, and as I just explained to 10:15:52 12 you, there are real dangers in assuming that the side effects 10:15:55 13 of one drug are the same as the side effects of another. 10:15:58 14 The second is they can't even establish that Prozac 10:16:02 15 causes suicide because their case reports -- you know, in my 10:16:06 16 diagram case reports can raise questions for you. They can 10:16:11 17 cause epidemiologists to raise questions and then go do 10:16:14 18 studies. Case reports by themselves don't allow you to cross 10:16:19 19 the river and actually answer whether in this case, you know, 10:16:26 20 they're about Prozac, whether Prozac actually causes 10:16:29 21 suicidality. 10:16:31 22 Q. Dr. Wang, to a reasonable degree of medical and scientific 10:16:34 23 certainty does Paxil cause suicides? 10:16:38 24 A. Please repeat that question. 10:16:39 25 Q. To a reasonable degree of medical and scientific certainty 1040 10:16:43 1 does Paxil cause suicide? 10:16:46 2 A. No. 10:16:46 3 Q. To a reasonable degree of medical and scientific certainty 10:16:49 4 does Paxil cause suicide in a subpopulation of people? 10:16:56 5 A. No, it does not. 10:17:00 6 Q. And to a reasonable degree of medical and scientific 10:17:02 7 certainty is there reasonable evidence of an association 10:17:06 8 between Paxil and suicide? 10:17:07 9 A. No, there is not. 10:17:09 10 Q. Do you claim that Paxil causes suicide, Doctor? Would 10:17:12 11 that be what you termed a wrong claim? 10:17:14 12 A. That would -- it is potentially a wrong claim of the type 10:17:19 13 that I sort of talked about earlier, and it is potentially -- 10:17:24 14 it could potentially be a problem. Again, if there's -- 10:17:28 15 there's already underuse of these psychiatric medications by 10:17:32 16 patients that really need them and would benefit from them. 10:17:35 17 If there are wrong claims made about these medications, there 10:17:39 18 will be even more underuse of these medications by people who 10:17:42 19 would benefit. People who do ultimately take the medications 10:17:47 20 will wait even longer. 10:17:49 21 Remember, I told you our data shows about 11 years 10:17:53 22 elapsing before people experiencing symptoms of depression or 10:17:58 23 other psychiatric illnesses and then actually treat 10:18:01 24 illnesses. Over a decade. That's a long time. You may even 10:18:06 25 see a worsening in it in people waiting longer because wrong 1041 10:18:11 1 claims are made. 10:18:12 2 Lastly, among people who do get treated, they may 10:18:16 3 needlessly worry. If there are a lot of claims made about, 10:18:19 4 you know, whether the antidepressant they're on causes things 10:18:23 5 like suicide, they will worry, potentially needlessly and 10:18:28 6 maybe come off the medication sooner than they really should. 10:18:32 7 MR. PREUSS: Thank you. No further questions at this 10:18:34 8 time. 10:18:34 9 THE COURT: Mr. Vickery. 10 CROSS-EXAMINATION 10:18:50 11 Q. (BY MR. VICKERY) Good morning, Dr. Wang. 10:18:52 12 A. Good morning, Mr. Vickery. 10:18:54 13 Q. You have brought a file with you to the stand. What is 10:18:56 14 that? 10:18:57 15 A. They're the studies which I referred to and also my 10:19:01 16 Rule 26 and also a copy of the diagram that I put up there. 10:19:08 17 Q. Did you select the articles to read yourself or were they 10:19:11 18 selected for you? 10:19:14 19 A. There was a mixture. You mean by selected, selected by 10:19:17 20 whom? 10:19:18 21 Q. By one of these lawyers. 10:19:21 22 A. They provided me with some articles and I also looked up 10:19:24 23 some articles myself, and certain articles reference other 10:19:28 24 studies and I looked them up myself. 10:19:30 25 Q. I note, for example, one of the things that's not listed 1042 10:19:33 1 on your report is a document that's the Consensus Statement 10:19:44 2 of the American College of Neuropsychopharmacologists in 10:19:44 3 1992. Are you familiar with that document? 10:19:46 4 A. The one written by John Mann? 10:19:52 5 Q. Yes. 10:19:52 6 A. Yes, I've read it. 10:19:54 7 Q. In connection with this case? 10:19:56 8 A. In connection with this case. 10:19:57 9 Q. Is it in your folder? 10:19:59 10 A. It is not in my folder. 10:20:01 11 Q. Did you read it at your own instance or because someone 10:20:05 12 suggested that you should? 10:20:06 13 A. It came up because it is part of the literature on this 10:20:08 14 case. I don't know if I looked it up or if someone else 10:20:11 15 looked it up. 10:20:12 16 Q. May I see your file there? 10:20:13 17 A. Sure. Actually, it may be in here. I need to look. 10:20:31 18 Q. I will have a look at the break and return it to you 10:20:35 19 afterwards. 10:20:36 20 Dr. Wang, the research you described on breast 10:20:38 21 cancer, who funded that? 10:20:40 22 A. That was funded -- the research was funded by the National 10:20:45 23 Institute of Mental Health. 10:20:45 24 Q. And what was the study design? 10:20:47 25 A. The study design, as I described, it is the type of study 1043 10:20:51 1 that fits in the -- my left-hand bottom corner. It is an 10:20:58 2 observational study. So there wasn't random assignment. But 10:21:02 3 it was adjusted for any prescribing bias, you know, whereby 10:21:08 4 antidepressants would be given to women who are at higher 10:21:12 5 risk of breast cancer. 10:21:13 6 Q. Was it a cohort study? 10:21:15 7 A. It was a cohort study. 10:21:17 8 Q. Would you tell the jury what a cohort study is. 10:21:21 9 A. A cohort study is a type of observational study where, 10:21:25 10 again, you look at how antidepressants are being used in the 10:21:28 11 real world. You -- no one is randomly assigning patients, it 10:21:32 12 is not an experiment, we just observed how antidepressants 10:21:36 13 were being used, actually in the state of New Jersey. 10:21:39 14 Q. Okay. Was it a prospective study or retrospective? 10:21:47 15 A. It was a retrospective study. 10:21:49 16 Q. Would you tell the jury the difference between prospective 10:21:52 17 and retrospective studies? 10:21:54 18 A. A retrospective -- you're talking about epidemiologic 19 studies? 10:21:59 20 Q. Yes, sir. 10:22:01 21 A. An epidemiologic study that is retrospective is a study 10:22:06 22 where the data that we used for the study was collected after 10:22:11 23 the exposures were given. In this case the exposures are 10:22:16 24 just the prescriptions, you know, that doctors are giving to 10:22:20 25 patients in the state of New Jersey. And it also -- we 1044 10:22:25 1 collected the data and did the analysis after the breast 10:22:29 2 cancer cases occurred. 10:22:31 3 So we collected the data on the exposure and also the 10:22:35 4 disease, in this case breast cancer -- we collected the data 10:22:39 5 and did the analyses after both of those things happened. 10:22:43 6 That's retrospective. 10:22:45 7 Q. Would you mind stepping down from the board and flipping 10:22:47 8 the chart, and as you've done with questions and answers, put 10:22:50 9 prospective on one side and retrospective on the other? 10:22:54 10 Would you do that for us? 10:22:56 11 MR. PREUSS: Your Honor, before he flips it, I would 10:22:59 12 like to move the exhibit into evidence. 10:23:01 13 THE COURT: Is this JJ? 10:23:02 14 MR. PREUSS: Yes, sir. 10:23:03 15 THE COURT: Any objection? 10:23:04 16 MR. VICKERY: No objection. 10:23:05 17 THE COURT: Defendant's Exhibit JJ may be received in 10:23:07 18 evidence. 10:23:09 19 (Defendant Exhibit JJ received in evidence.) 10:23:14 20 THE WITNESS: What does that mean? 10:23:19 21 MR. VICKERY: It means the jury gets it at the end of 10:23:21 22 the case. 10:23:22 23 Q. (BY MR. VICKERY) Would you mind flipping it over and 10:23:23 24 write for us prospective and retrospective. 10:23:40 25 Would you be so kind as to draw a body of water 1045 10:23:43 1 between those two, a stream or river or something. 10:23:48 2 Now, in the overall pecking order of scientific 10:23:52 3 validity from an epidemiologist's standpoint are prospective 10:23:58 4 studies generally regarded as more scientifically rigorous 10:24:02 5 than retrospective studies? 10:24:06 6 A. Let's get some terms clear here, if you don't mind. 10:24:13 7 Q. Let me rephrase it. 10:24:15 8 A. Yeah, because I'm not sure -- 10:24:18 9 Q. I don't want to use buzzwords, just plain English. 10:24:20 10 Are prospective studies better than retrospective 10:24:26 11 studies at answering questions? 10:24:27 12 A. If you mean is it better to collect the data -- is it 10:24:32 13 better to collect the data before the exposure occurred and 10:24:37 14 before the disease occurred, then yes, that's better than 10:24:40 15 collecting the data after the exposure occurred and the 10:24:48 16 disease occurred. 10:24:49 17 Q. So prospective is better? 10:24:52 18 A. As long as we're understanding each other. I want to make 10:24:56 19 sure we're using the same terms. 10:25:03 20 Q. I think we are. 10:25:03 21 I mean, are you aware, for example, that way back in 10:25:03 22 1990 Dr. David Wheadon of SmithKline Beecham went to Europe 10:25:06 23 and met with seven international experts, opinion leaders on 10:25:10 24 this issue of suicide, and at that time Prozac was what he 10:25:14 25 was focusing on, and they unanimously said the best way to 1046 10:25:18 1 study this is with a prospective study? Are you aware of 10:25:27 2 that? 10:25:27 3 A. Prospective data? 10:25:28 4 Q. Prospective study? 10:25:29 5 A. Prospectively collected data? 10:25:32 6 Q. Yes, sir. 10:25:33 7 A. I'm not aware of what you're talking about. 10:25:36 8 Q. No one has told you that? 10:25:37 9 A. No, I have no idea what you're talking about. 10:25:39 10 Q. Would you agree with those seven international experts 10:25:42 11 that if we wanted to find out whether Paxil poses a risk of 10:25:47 12 violence or suicide for a small, vulnerable subpopulation of 10:25:52 13 people, that the best way to answer that question is with a 10:25:55 14 prospective study rather than a retrospective study? 10:25:59 15 A. You would want to prospectively -- you would want to 10:26:05 16 collect your data before. You would want to enroll your 10:26:12 17 patients in your study before the disease -- before the 10:26:17 18 exposures have happened and before the disease has occurred. 10:26:20 19 Q. So are you saying you agree with the seven international 10:26:23 20 experts that it is better to study it prospectively? 10:26:26 21 A. You would want to, yeah, enroll your patients in the 10:26:29 22 study, you would want to have planned your study, yeah, 10:26:33 23 before. 10:26:33 24 Q. Now, study design is very, very important for an 10:26:37 25 epidemiologist, isn't it? 1047 10:26:41 1 A. Potentially, yes. 10:26:42 2 Q. I mean, I was noting the metaphor you used, and I think it 10:26:46 3 is a good one, you say the study has to be properly equipped 10:26:50 4 and has to have a skilled pilot like the airplane, right? 10:26:54 5 A. Right. 10:26:55 6 Q. And when you're talking about a properly equipped study, 10:26:58 7 you're talking about one that has clearly defined goals from 10:27:02 8 the outset that is designed to answer the question. True or 10:27:06 9 not true? 10:27:07 10 A. Repeat that question. 10:27:09 11 Q. In the field of epidemiology when one is going to conduct 10:27:11 12 a prospective study, is it important to design that study so 10:27:15 13 that it is, you know, honed in and focused on the question 10:27:20 14 you're trying to answer? 10:27:21 15 A. Not necessarily, no. 10:27:23 16 Q. No? 10:27:24 17 A. No. 10:27:25 18 Q. So is there something in randomized clinical trials called 10:27:30 19 the stated hypothesis? 10:27:33 20 A. There are times when you, for example, might want to do 10:27:37 21 your analyses afterwards. That may be the best thing and the 10:27:40 22 right thing to do. We may be disagreeing on -- may not be 10:27:44 23 using the terms in the same way. 10:27:46 24 Q. I'm talking about study design in the field of 10:27:48 25 epidemiology. 1048 10:27:50 1 A. Right. 10:27:50 2 Q. Is it important when you're going to conduct a study to 10:27:53 3 know what questions you're going to answer? 10:27:57 4 A. No, not necessarily. 10:28:00 5 Q. So like, to take your example, if you had done a 10:28:06 6 randomized clinical trial, a controlled trial, that's 10:28:11 7 prospective, right? 10:28:12 8 A. That would be prospective. 10:28:13 9 Q. And if you had done one to find out whether Prozac causes 10:28:18 10 breast cancer, you would have stated that in the hypothesis 10:28:22 11 of the study, wouldn't you? 10:28:24 12 A. No, there's -- I slipped earlier when I was saying -- and 10:28:29 13 let me expand a little bit on it. 10:28:32 14 There is actually -- the first thing we actually 10:28:34 15 looked for was was there existing clinical trial data that we 10:28:39 16 could examine. And the reason why that would be the first 10:28:41 17 and best thing to do is for the following reason: If we had 10:28:47 18 waited, tried to go out and do clinical trials, we would have 10:28:51 19 had to wait. Do you know how long it takes for breast cancer 10:28:55 20 to develop? 10:28:56 21 Q. I do. 10:28:57 22 A. So actually the best and first thing you should do is look 10:29:00 23 at randomized controlled clinical trial data existing for the 10:29:07 24 following reason: One is if you wait and actually do a study 10:29:09 25 where you, you know, randomized women to antidepressant, 1049 10:29:13 1 nonantidepressant, placebo, you potentially would have to 10:29:17 2 wait 30 or more years before you can answer the question. 10:29:21 3 And that may be too long. It is also expensive to do that. 10:29:26 4 Q. People could die while you're doing the study? 10:29:28 5 A. While you're doing the study. 10:29:30 6 And the other reason why you would seek out existing 10:29:32 7 clinical trial data, the reason you would seek it out first 10:29:35 8 is because it is the highest quality data. 10:29:37 9 So the first thing you do is try to get data, 10:29:40 10 existing data for a rapid answer that's from that -- you 10:29:43 11 know, my left-hand top-hand corner. 10:29:47 12 Q. And was there such data? 10:29:55 13 A. On what? 10:29:55 14 Q. The existing clinical trials. 10:29:55 15 A. Are we talking antidepressants and breast cancer? 10:29:56 16 Q. Yes, sir. Had anyone collected data when they did the 10:29:58 17 existing randomized controlled trials that was designed to 10:30:02 18 answer the question of whether these drugs cause breast 10:30:05 19 cancer? 10:30:05 20 A. Design is not the important issue. It is -- we needed to 10:30:09 21 see, we needed to find was there existing clinical trial 10:30:14 22 data. If it had been available, that would have been our 10:30:16 23 first and preferred means using the already collected data 10:30:19 24 from existing clinical trials. 10:30:21 25 Q. But it wasn't there, was it? 1050 10:30:23 1 A. The trials -- the randomized clinical trials didn't exist. 10:30:27 2 Q. There had been randomized clinical trials with those 10:30:30 3 drugs, hadn't there? 10:30:31 4 A. Not where they collected breast cancer -- 10:30:34 5 Q. My very point. The trials that had been conducted -- 10:30:38 6 there had been umpteen zillion randomized clinical trials 10:30:43 7 with Prozac, hadn't there, Dr. Wang? 10:30:47 8 A. What do you mean by umpteen? 10:30:49 9 Q. A whole bunch, there had been a whole bunch? 10:30:52 10 A. To study breast cancer you need a lot because it is a 10:30:55 11 rare -- it is not that rare, but it is rare in the sense that 10:30:57 12 among, for example, especially a young group of women you 10:31:01 13 have to have a lot and you have to wait a long time before 10:31:03 14 they develop breast cancer so you need a lot. When you say 10:31:08 15 umpteen, even though there may have been 20, 30, you would 10:31:13 16 still might even need more than that. 10:31:16 17 In any case, the trials weren't available for the 10:31:21 18 study. 10:31:22 19 Q. When you started looking into that issue did you have some 10:31:24 20 idea of the incidence rate, if it were true, that Prozac 10:31:27 21 caused breast cancer? Did you have in your mind from the 10:31:31 22 case reports or animal studies some notion of what -- of how 10:31:35 23 rare it would be, to use your word rare? 10:31:40 24 A. You mean what proportion of women ultimately get breast 10:31:43 25 cancer? 1051 10:31:44 1 Q. Yes, sir. 10:31:44 2 A. A figure is about maybe 1 in 9, 1 in 10 women ultimately 10:31:49 3 in their lives get breast cancer. 10:31:51 4 Q. I'm talking about what you expected the incidence rate to 10:31:54 5 be for those women who were on the drug. You said it was a 10:31:57 6 rare event. What did you mean by rare? 10:32:02 7 A. What I meant by rare is in -- if you examine particularly 10:32:08 8 a younger population, starting out with a population that has 10:32:11 9 a bunch of younger women in it, breast cancer might be rare, 10:32:16 10 let's say, for example, if you're studying a population with 10:32:20 11 a lot of 20- or 30-year-olds. They aren't the ones that get 10:32:24 12 breast cancer. Unfortunately it is women in the later 10:32:29 13 decades. 10:32:30 14 Q. Are we talking 2 or 3 percent of the population? Is that 10:32:32 15 rare? 10:32:33 16 A. Let me just clarify, in the entire population -- in other 10:32:37 17 words, if you look at women who at the end of their lives -- 10:32:41 18 if you're only looking at women who have completed their 10:32:43 19 lives, maybe about 1 in 9 women in the world will develop 10:32:47 20 breast cancer. 10:32:48 21 But, it is different to do a study that includes 10:32:52 22 women in their 20s, 30s, 40s. It is going to be rarer 10:32:57 23 because these women haven't gotten to the end of their lives, 10:33:00 24 so breast cancer will be rarer. 10:33:02 25 Q. Dr. Wang, was your study designed to detect a doubling of 1052 10:33:06 1 the risk? 10:33:07 2 A. Yes, it was. 10:33:12 3 Q. Odds ratio, relative risk, what was the measure? 10:33:16 4 A. We used a variety of measures, one called a hazard ratio. 10:33:20 5 It is from, just to give you a term, survival analysis. The 10:33:25 6 reason we use that is because of the dropout rate. Women can 10:33:30 7 drop out of our study, so you have to use survival analysis. 10:33:34 8 Q. And survival analysis is the mathematical way to take the 10:33:38 9 few women left and try to extrapolate from that data to cover 10:33:42 10 the whole patient base; is that right? 10:33:45 11 A. If you have dropouts -- if women sort of move out of the 10:33:49 12 state of New Jersey, we lose information on them, you know, 10:33:53 13 in our study, and because of that kind of dropout or loss of 10:33:58 14 information we use survival analysis. 10:34:00 15 Q. And was your study specifically designed to detect a 10:34:04 16 doubling of the risk? 10:34:06 17 A. It was powered to do that. 10:34:10 18 Q. Powered to do that, I think I may know because I've tried 10:34:13 19 to learn a little epidemiology, but would you explain to 10:34:17 20 these folks what it means when you say it was powered to do 10:34:21 21 that? 10:34:21 22 A. Sure. If I can refer back to my -- remember when I 10:34:24 23 defined what statistical significance means? The -- what 10:34:30 24 powered means is you could show statistical significance. 10:34:36 25 And just, if I may review, statistical significance when you 1053 10:34:40 1 see a difference, like in our case, let's say, for example, 10:34:43 2 we saw -- I'll just go back to my example, if here's patients 10:34:51 3 who are taking drug A, here are patients taking drug B or 10:34:55 4 nothing or sugar pill, you may find a difference on whatever 10:34:58 5 measure you're looking at. Numerically if you look at just 10:35:02 6 the numbers, this percentage may be higher than this 10:35:05 7 percentage. 10:35:06 8 And you need to say, "Is this a real difference. Is 10:35:12 9 it just due to chance or error" -- 10:35:14 10 Q. Can I try -- I'm sorry. I didn't mean to interrupt you. 10:35:18 11 See if this will help: Are you saying when it is powered 10:35:21 12 that you need enough women in the study to detect a real 10:35:25 13 phenomenon? 10:35:27 14 A. To show a statistically significant finding if it is 10:35:32 15 there. If it is there. 10:35:33 16 Q. How many woman would you have needed to answer that 10:35:36 17 question for this rare event? 10:35:37 18 A. We did a calculation. I can't off the top of my head 10:35:41 19 actually tell you how many. 10:35:42 20 Q. Ballpark. Thousands? 10:35:45 21 A. Thousands. 10:35:47 22 Q. Tens of thousands? 10:35:48 23 A. Tens of thousands. 10:35:49 24 Q. So you would need a large, large study population to get a 10:35:52 25 reliable, statistically significant answer to that question, 1054 10:35:56 1 right? 10:35:57 2 A. We didn't -- there are different levels, also, of 10:36:00 3 significance. And what -- let me sort of try to simply say, 10:36:08 4 to show a difference it depends how big a difference you want 10:36:11 5 to show. If you want to show a very, very big difference, 10:36:15 6 you don't need that many people. If you want to show a very, 10:36:18 7 very small difference is a real one, you need more patients. 10:36:23 8 Q. If it is a rare phenomenon, if it involves a small 10:36:26 9 percentage of patients, you need a lot more people to power 10:36:30 10 the study properly, don't you? 10:36:33 11 A. Potentially. 10:36:34 12 Q. Okay. 10:36:35 13 THE COURT: Why don't we take our break now, 10:36:38 14 Mr. Preuss? 10:36:39 15 MR. PREUSS: That would be fine. 10:36:40 16 THE COURT: Take our morning recess and stand in 10:36:42 17 recess for 15 minutes. 10:36:45 18 (Recess taken 10:35 a.m. until 10:55 a.m.) 10:59:29 19 THE COURT: Before we proceed, Dr. Wang, you 10:59:32 20 understand you're still under oath? 10:59:34 21 THE WITNESS: Yes, I do. 10:59:36 22 Q. (BY MR. VICKERY) Dr. Wang, did you receive any 10:59:38 23 information or instructions during the break that will assist 10:59:41 24 you in answering my questions this morning? 10:59:44 25 A. Could you repeat the question? 1055 10:59:46 1 Q. Did you get any information or anything during the course 10:59:48 2 of the break that alters any of the answers you've given so 10:59:51 3 far or affects any of the answers you're going to give in the 10:59:54 4 future? 10:59:55 5 A. No. 10:59:56 6 Q. Okay. We were talking before the break about prospective 11:00:01 7 and retrospective studies and specifically about study 11:00:05 8 design. Would you agree, to continue with your airplane 11:00:24 9 metaphor, that it is appropriate and important that you have 11:00:28 10 rating instruments that are adequate to the task; in other 11:00:32 11 words, that will measure what it is you need to look at? 11:00:35 12 A. Yes, it is important to have -- by things that you're 11:00:39 13 measuring, I think you mean outcomes and exposures. And yes, 11:00:44 14 it is important to have good measures of exposures to things 11:00:47 15 like drugs and also good measures of the outcomes you're 11:00:51 16 looking for. 11:00:51 17 Q. So if the outcome we're looking for is treatment-emergent 11:00:57 18 suicidality, it is important to use a rating scale that 11:01:01 19 measures that in an accurate way, isn't it? 11:01:05 20 A. Yes, that's a good thing. 11:01:06 21 Q. You said you read Dr. Mann's paper he wrote on behalf of 11:01:11 22 the ACNP. Did you also read his paper that he wrote in 1991 11:01:15 23 with his colleague Dr. Kapur, the Mann and Kapur -- 11:01:19 24 A. You said did I write it? 11:01:22 25 Q. Read it. 1056 11:01:23 1 A. Did I read it? 11:01:24 2 Q. Yes. 11:01:24 3 A. I don't think I read that. If I did, it escapes me. I 11:01:27 4 don't recall it. 11:01:28 5 Q. All right. You're aware of who Dr. J. John Mann is, 11:01:32 6 aren't you? 11:01:33 7 A. I only know that he's an expert. I actually don't know 11:01:36 8 him personally. 11:01:38 9 Q. Are you aware of his reputation in the field of 11:01:40 10 suicidology? 11:01:41 11 A. I only know that he is. I'm not a suicidologist so I 11:01:46 12 don't know his reputation in particular. 11:01:48 13 Q. You're not a suicidologist or a psychopharmacologist, are 11:01:53 14 you? 11:01:54 15 A. Could you define psychopharmacologist? 11:01:57 16 Q. Well, would you call yourself one? 11:02:00 17 A. It depends what you're trying to -- what you mean by that. 11:02:04 18 In other words, somebody who prescribes drugs, researches 11:02:08 19 psychiatric medication -- 11:02:09 20 Q. Would you call yourself a psychopharmacologist? 11:02:12 21 A. I do research on psychiatric medications. 11:02:15 22 Q. Are you a member of the American College of 11:02:18 23 Neuropsychopharmacologists? 11:02:21 24 A. Psychopharmacology. 11:02:23 25 Q. The American College of Neuropsychopharmacology? 1057 11:02:30 1 A. I'm not a member. You would actually have to be selected 11:02:31 2 into it. But I did attend the annual meeting that's held for 11:02:34 3 this society last December. 11:02:37 4 Q. So are you a psychopharmacologist or not? 11:02:39 5 A. I do research on psychiatric medications. 11:02:43 6 Q. Don't you tell people that you are a 11:02:45 7 pharmacoepidemiologist? 11:02:47 8 A. That I am. 11:02:48 9 Q. And the department you're in, incidentally, is the 11:02:54 10 Department of Pharmacoepidemiology and Pharmacoeconomics, 11:02:58 11 isn't it? 11:02:59 12 A. Division of, yes. 11:03:00 13 Q. And what is pharmacoeconomics? 11:03:03 14 A. Pharmacoeconomics is also a discipline that's becoming 11:03:08 15 increasingly important because it is -- for better or worse, 11:03:17 16 it is not just how much drugs work or whether they're safe. 11:03:21 17 That's clearly something important to study. But 11:03:23 18 increasingly since health care expenditures are so high it is 11:03:28 19 important to know if a drug is beneficial, how much does it 11:03:31 20 cost for that benefit, you know. 11:03:33 21 So these are specific types of scientific studies 11:03:35 22 where you not only measure the effects of the drug in terms 11:03:40 23 of clinical terms, but you also measure them in terms of 11:03:44 24 their costs, maybe cost savings. 11:03:47 25 Q. In the practice guideline you wrote for depression did you 1058 11:03:51 1 suggest that the best therapy is a combination of drug 11:03:57 2 therapy on the one hand plus traditional psychotherapy on the 11:04:01 3 other? 11:04:01 4 A. I didn't write the guidelines. I was a consultant to the 11:04:04 5 work group that prepared them. 11:04:06 6 Q. Do the guidelines say that the best therapy is the 11:04:10 7 combination of psychotherapy and, where indicated, drug 11:04:15 8 therapy together? 11:04:16 9 A. Are you referring to the -- what are you referring to, by 11:04:18 10 the way, which guideline? 11:04:20 11 Q. The one on depression. 11:04:22 12 A. Among certain patients the combination may be better. 11:04:25 13 Q. And you're aware of published literature that found to a 11:04:28 14 statistically significant degree with an appropriate P value 11:04:31 15 and convergence interval that that is indeed the case? 11:04:36 16 You're aware of that literature, aren't you? 11:04:38 17 A. I reviewed that literature a while ago and, again, I'm not 11:04:42 18 an expert in combination modality treatment. By combination 11:04:47 19 modality it means giving a medication and also psychotherapy 11:04:52 20 together. That's not my area of expertise. I did read the 11:04:56 21 articles when I was a member of the work group, consultant to 11:04:59 22 the work group, and they're not fresh in my mind so I can't 11:05:02 23 really discuss them. 11:05:03 24 Q. From a pharmacoeconomic standpoint it is sure cheaper to 11:05:07 25 just give a guy a pill than to spend an hour of a doctor's 1059 11:05:11 1 time with him each week getting to know him and talking, 11:05:16 2 helping him talk through his problems? Cheaper, isn't it, to 11:05:20 3 give the pill alone? 11:05:22 4 A. No, that's not true. That's where pharmacoeconomics is so 11:05:25 5 important. That may be a hypothesis. You may have that as a 11:05:29 6 question that you asked, but you then actually go and do the 11:05:33 7 studies where you actually need to evaluate. The answer 11:05:37 8 isn't clear to that question you just raised. 11:05:39 9 Q. Why not? I mean, somebody has got to pay for the hour of 11:05:42 10 that doctor's time, don't they? 11:05:46 11 A. It is not clear. It has a lot to do with you need good 11:05:51 12 data from randomized clinical trials in order to do these 11:05:54 13 studies in which you need to know how big an effect is -- how 11:05:58 14 big is the beneficial effect of psychotherapy, talking to 11:06:02 15 someone. You also need to know how big the positive effect 11:06:06 16 is of antidepressants. You need to know what side effects 11:06:10 17 they might cause because each of those can have costs. And 11:06:13 18 you need to know how much money is involved in each of these 11:06:16 19 treatments. 11:06:17 20 You need all of this information. And again, ideally 11:06:20 21 your information, if you can get it, would come from 11:06:23 22 randomized controlled clinical trials. 11:06:26 23 Q. Dr. Wang, as part of your residency you were trained to do 11:06:29 24 psychotherapy, weren't you? 11:06:31 25 A. I received training in it as part of a general adult 1060 11:06:34 1 psychiatry residency. 11:06:36 2 Q. And your rate is $300 an hour, correct? 11:06:39 3 A. For working with -- as a consultant to the lawyers. 11:06:46 4 Q. Isn't it cheaper to give someone a Paxil pill than to 11:06:49 5 spend that hour in psychoth