1 NO. 90-CI-6033 JEFFERSON CIRCUIT COURT DIVISION ONE (1) 2 3 JOYCE FENTRESS, ET AL. PLAINTIFFS 4 5 VS. DEPOSITION FOR PLAINTIFFS 6 7 SHEA COMMUNICATIONS, ET AL. DEFENDANTS 8 * * * * * * * * * * 9 10 DEPONENT: GREGORY ENAS, PH.D 11 DATE: SEPTEMBER 16 AND 17, 1993 12 13 * * * * * * * * * * 14 15 16 REPORTER: KATHY NOLD 17 18 KENTUCKIANA REPORTERS SUITE 260 19 730 WEST MAIN STREET LOUISVILLE, KENTUCKY 40202 20 (502) 589-2273 Page 1 1 * * * * * * * * * * 2 3 UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF INDIANA 4 INDIANAPOLIS DIVISION 5 IN RE ELI LILLY AND COMPANY ) Prozac Products Liability ) MDL Docket No. 907 6 Litigation ) 7 * * * * * * * * * * 8 NO. 91-02496-A 9 JACKIE LYNN BIFFLE, ET AL ) IN THE DISTRICT ) COURT OF 10 V. ) DALLAS COUNTY, TEXAS ) 11 ELI LILLY & COMPANY AND ) 14TH JUDICIAL DISTA PRODUCTS COMPANY ) DISTRICT 12 * * * * * * * * * * 13 NO. 92-14775-E 14 RICHARD HAROLD CROSSETT, JR., ) IN THE 15 CHAD H. CROSSETT, AMY MICHELLE ) DISTRICT CROSSETT AND KRISTEN ANN CROSSETT, ) COURT OF 16 INDIVIDUALLY AND AS SURVIVORS OF ) AND ON BEHALF OF THE ESTATE OF ) 17 JOCQUETTA ANN CROSSETT, DECEASED ) ) 18 V. ) DALLAS COUNTY, ) TEXAS 19 ELI LILLY & COMPANY, DISTA ) PRODUCTS COMPANY, TEXAS ) 20 PSYCHIATRIC COMPANY, INC. ) D/B/A/ HCA WILLOW PARK ) 101ST JUDICIAL 21 HOSPITAL, JAMES K. WITSCHY, M.D., ) DISTRICT AND DOUG BELLAMY, ED.D. ) Page 2 1 * * * * * * * * * * 2 NO. A-921,405-C 3 MARIA GUADALUPE REVES ) IN THE 4 INDIVIDUALLY AND AS NEXT ) DISTRICT COURT FRIEND OF GRANT JULIAN REVES ) OF 5 A MINOR CHILD, AND ON BEHALF ) OF THE ESTATE OF CHRISTIAN ) 6 MARIE REVES, DECEASED ) ) ORANGE COUNTY, 7 V. ) TEXAS ) 8 ELI LILLY & COMPANY, DISTA ) PRODUCTS COMPANY, RAVIKUMAR ) 9 KANNEGANTI, M.D., HOSPITAL ) CORPORATION OF AMERICA, A ) 10 TENNESSEE CORPORATION, HEALTH ) SERVICES ACQUISITION CORP., ) 11 A DELAWARE CORPORATION, ) HCA PSYCHIATRIC COMPANY, A ) 12 DELAWARE CORPORATION, TEXAS ) PSYCHIATRIC CO., INC.. A/K/A ) 13 AND/OR D/B/A HCA BEAUMONT ) NEUROLOGICAL HOSPITAL, AND HCA ) 14 HEALTH SERVICES OF TEXAS, INC. ) 128TH JUDICIAL A/K/A AND/OR BEAUMONT ) DISTRICT 15 NEUROLOGICAL HOSPITAL ) 16 * * * * * * * * * * Page 3 1 IN THE UNITED STATES DISTRICT COURT 2 FOR THE WESTERN DISTRICT OF TEXAS SAN ANTONIO DIVISION 3 ELIZABETH T. SANCHEZ, ) 4 INDIVIDUALLY AND AS THE ) SURVIVING SPOUSE, MARGARET R. ) 5 SANCHEZ, INDIVIDUALLY AND NEXT ) OF FRIEND OF DEBRA JEAN ) 6 SANCHEZ, VERONICA MARIE ) SANCHEZ, EDWARDO ESTEBAN ) 7 SANCHEZ, AND MICHAEL ANTHONY ) SANCHEZ, CHILDREN; AND ALL ON ) 8 BEHALF OF THE ESTATE OF ) EDWARDO SANCHEZ ) 9 ) V. ) CIVIL ACTION NO. 10 ) SA93CA367 ELI LILLY AND COMPANY AND ) 11 DISTA PRODUCTS COMPANY ) 12 * * * * * * * * * * 13 IN THE UNITED STATES DISTRICT COURT FOR THE SOUTHERN DISTRICT OF TEXAS 14 HOUSTON DIVISION 15 MARIA SANCHEZ, INDIVIDUALLY ) AND AS NEXT FRIEND OF DEBORAH ) 16 SANCHEZ, VERONICA SANCHEZ, ) EDDIE SANCHEZ, AND MICHAEL ) 17 SANCHEZ, AND ON BEHALF OF THE ) ESTATE OF EDUARDO SANCHEZ ) 18 ) V. ) CIVIL ACTION NO. 19 ) H-93-1469 ELI LILLY AND COMPANY AND ) 20 DISTA PRODUCTS COMPANY, A ) DIVISION OF ELI LILLY AND ) 21 COMPANY ) Page 4 1 * * * * * * * * * * 2 STATE OF NEW YORK 3 SUPREME COURT COUNTY OF JEFFERSON 4 _____________________________________________ 5 STEPHANIE CAPONE, AS EXECUTOR OF THE ESTATE OF JOSEPH J. CAPONE, JR., AND 6 STEPHANIE CAPONE, INDIVIDUALL, NOTICE TO TAKE 7 PLAINTIFF, DEPOSITION UPON ORAL EXAMINATION 8 VS. INDEX NO. 93-251 9 ELI LILLY AND COMPANY, DISTA PRODUCTS 10 COMPANY, A DIVISION OF ELI LILLY AND COMPANY, FLOYD BAJJALY, M.D, 11 DEFENDANTS. 12 _____________________________________________ 13 * * * * * * * * * * 14 SUPREME COURT OF TEH STATE OF NEW YORK COUNTY OF ORANGE 15 --------------------------------------X BRUCE R. MALEN AS EXECUTOR OF THE : INDEX NO. 16 ESTATE OF BARBARA E. MALEN, AND OF : 4119/92 BRUCE R. MALEN, INDIVIDUALLY, : 17 : HON. PETER PLAINTIFF : PATSALOS, 18 : J.S.C. -against- : 19 : ELI LILLY & COMPANY, DISTA PRODUCTS : 20 COMPANY, A DIVISION OF ELI LILLY & : COMPANY, BARRY SINGER AND UNITED : 21 HOSPITAL, : : 22 DEFENDANTS. : --------------------------------------X 23 * * * * * * * * * * Page 5 1 ---------------------------------X 2 VALARIE J. FRIEDMAN AND DAVID : SUPERIOR COURT FRIEDMAN, HER HUSBAND, : OF NEW JERSEY 3 : LAW DIVISION: PLAINTIFF, : MIDDLESEX COUNTY 4 : DOCKET NO. : L-3191-91 5 VS. : : 6 ELI LILLY & COMPANY; DISTA : PRODUCTS INC, A DIVISION OF : 7 ELI LILLY & COMPANY; LISS : PHARMACY; MADISON PHARMACY AND : 8 JOHN DOES NOS. 1-25 (UNKNOWN : ENTITIES), : 9 : DEFENDANTS. : 10 ---------------------------------X 11 * * * * * * * * * * 12 SUPREME COURT OF THE STAET OF NEW YORK COUNTY OF SUFFOLK 13 -------------------------------------x 14 RHOMDA L. HALA and JOSEPH L. HALA, : 15 Plaintiffs, : Index No. 14869/90 16 - against - : 17 ELI LILLY & COMPANY and DISTA : PRODUCTS COMPANY, a DIVISION OF 18 ELI LILLY & COMPANY : 19 Defendants. : -------------------------------------x 20 21 * * * * * * * * * * Page 6 1 IN THE CIRCUIT COURT OF COOK COUNTY, ILLINOIS 2 COUNTY DEPARTMENT, LAW DIVISION 3 PATRICIA BRACH, ) ) 4 Plaintiff ) ) 5 v. )No. 92 L 13369 ) 6 ELI LILLY AND COMPANY, a foreign ) corporation; ALAN N. MILLER, M.D., ) 7 WILLIAM BRUINSMA, Psy.D., and ) CONDELL MEMORIAL HOSPITAL, ) 8 ) Defendants. ) 9 * * * * * * * * * * 10 IN THE CIRCUIT COURT OF COOK COUNTY, ILLINOIS 11 COUNTY DEPARTMENT - LAW DIVISION 12 RENATO DI SILVESTRO, Individually ) and as Special Administrator of ) 13 the Estate of JOHN DI SILVESTRO, ) Deceased, ) 14 ) Plaintiff, ) 15 ) v. ) No. 91 L 7881 16 ) ROBERT L. NELSON, et al., ) 17 ) Defendants, ) 18 ) GEORGE MELNICK, M.D. and PETER ) 19 FINK, M.D. ) ) 20 Respondents in Discovery.) 21 * * * * * * * * * * Page 7 1 IN THE CIRCUIT COURT OF COOK COUNTY, ILLINOIS 2 COUNTY DEPARTMENT, LAW DIVISION 3 JOAN M. GRYER, ) ) 4 Plaintiff, ) ) 5 v. ) No. 92 L 7387 ) 6 ELI LILLY AND COMPANY, et al., ) ) 7 Defendants. ) 8 * * * * * * * * * * 9 IN THE CIRCUIT COURT OF COOK COUNTY, ILLINOIS 10 COUNTY DEPARTMENT, LAW DIVISION 11 JENNIFER HAMMERLI, as Plenary ) Guardian of the Estate of RAY B. ) 12 HAMMERLI, a disabled person, ) ) 13 Plaintiff, ) ) 14 v. ) No. 92 L 2365 ) 15 ELI LILLY AND COMPANY, THE ) UPJOHN COMPANY, DICKIE KAY, M.D., ) 16 (former Respondent in Discovery), ) and RICHARD CZECHOWICZ (former ) 17 Respondent in Discovery), ) ) 18 Defendants. ) 19 * * * * * * * * * * Page 8 1 IN THE CIRCUIT COURT OF THE SIXTH JUDICIAL CIRCUIT 2 CHAMPAIGN COUNTY, ILLINOIS 3 LINDA GARDNER, Individually and ) as Special Administrator of ) 4 the Estate of SHANE GARDNER, ) deceased, ) 5 ) Plaintiff, ) 6 ) v. ) No. 91 L 1066 7 ) ELI LILLY AND COMPANY, a foreign ) 8 corporation, ) ) 9 Defendant. ) 10 * * * * * * * * * * 11 IN THE NINETEENTH JUDICIAL CIRCUIT COURT 12 LAKE COUNTY, ILLINOIS 13 JAMES E. SHEPPARD, Special ) Administrator of the Estate of ) 14 KENNETH K. SHEPPARD, Deceased, ) ) 15 Plaintiff ) ) 16 v. ) No. 93 L 124 ) 17 GOOD SHEPHERD HOSPITAL, a ) corporation, DR. STEWART SEGAL, ) 18 DR. SANFORD SHERMAN, DR. BRUCE ) CARLSON, DR. R. BERGLUND, and ELI ) 19 LILLY & COMPANY, a corporation, ) ) 20 Defendants. ) 21 * * * * * * * * * * Page 9 1 SUPERIOR COURT OF THE STATE OF CALIFORNIA 2 FOR THE COUNTY OF LOS ANGELES 3 DR. MARIUS SAINES, etc., et al., ) Case No: 4 ) SC 008331 Plaintiffs, ) 5 ) vs. ) 6 ) ELI LILLY & COMPANY, a corporation; ) 7 DISTA PRODUCTS COMPANY, a division ) of Eli Lilly & Company; and DOBS 1- ) 8 100, inclusive, ) ) 9 Defendants. ) ____________________________________) 10 11 * * * * * * * * * * Page 10 1 THE DEPOSITION OF GREGORY ENAS, PH.D, TAKEN 2 AT THE OFFICE OF BAKER & DANIELS, 300 NORTH 3 MERIDIAN STREET, SUITE 2700, INDIANAPOLIS, 4 INDIANA 46204, ON SEPTEMBER 16 AND 17, 1993; SAID 5 DEPOSITION TAKEN PURSUANT TO NOTICE IN ACCORDANCE 6 WITH THE RULES OF CIVIL PROCEDURE. 7 * * * * * * * * * * 8 A P P E A R A N C E S 9 10 NANCY ZETTLER COUNSEL FOR GROUP A PLAINTIFFS 11 LEONARD M. RING AND ASSOCIATES, P.C. 111 WEST WASHINGTON AVENUE, SUITE 1333 12 CHICAGO, ILLINOIS 60602 13 LAWRENCE J. MYERS COUNSEL FOR ELI LILLY AND COMPANY 14 FREEMAN & HAWKINS 4000 ONE PEACHTREE CENTER 15 303 PEACHTREE STREET, N.E. ATLANTA, GEORGIA 30308-3243 16 MARGARET M. HUFF 17 ELI LILLY AND COMPANY LILLY CORPORATE CENTER 18 INDIANAPOLIS, INDIANA 46285 19 DENISE BRODSKY COUNSEL FOR GOOD SHEPHERD HOSPITAL 20 415 WASHINGTON STREET, SUITE 214 WAUKEGAN, ILLINOIS Page 11 1 MIGUEL A. RUIZ COUNSEL FOR DEFENDANTS CZECHOWICZ, FINK, BRUINSMA 2 CLAUSEN MILLER GORMAN CAFFREY & WITOUS 10 SOUTH LASALLE 3 CHICAGO, ILLINOIS 60603 4 PAUL J. CLEMENTI COUNSEL FOR DR. DICKIE KAY 5 HINSHAW & CULBERTSON 222 NORTH LA SALLE STREET, SUITE 300 6 CHICAGO, ILLINOIS 60601-1081 7 ROBIN HILL COUNSEL FOR DRS. WITSCHY AND KANNEGANTI 8 BAILEY AND WILLIAMS 3500 NCNB PLAZA 9 901 MAIN STREET DALLAS, TEXAS 75202-3714 10 PAUL SMITH 11 COUNSEL FOR PLAINTIFFS 745 CAMPBELL CENTER 2 12 8115 NORTH CENTRAL EXPRESSWAY DALLAS, TEXAS 75206 Page 12 1 I N D E X 2 3 DEPOSITION OF GREGORY ENAS, PH.D 4 5 DIRECT EXAMINATION BY MR. SMITH 14 6 CROSS EXAMINATION BY MS. ZETTLER 125 7 CROSS EXAMINATION BY MR. CLEMENTI 384 8 9 CERTIFICATE 394 10 ERRATA 395 11 12 EXHIBITS 13 PLAINTIFFS' EXHIBIT NO. 1 129 PLAINTIFFS' EXHIBIT NO. 2 139 14 PLAINTIFFS' EXHIBIT NO. 3 173 PLAINTIFFS' EXHIBIT NO. 4 195 15 PLAINTIFFS' EXHIBIT NO. 5 200 PLAINTIFFS' EXHIBIT NO. 6 242 16 PLAINTIFFS' EXHIBIT NO. 7 266 PLAINTIFFS' EXHIBIT NO. 8 268 17 PLAINTIFFS' EXHIBIT NO. 9 272 PLAINTIFFS' EXHIBIT NO. 10 275 18 PLAINTIFFS' EXHIBIT NO. 11 286 PLAINTIFFS' EXHIBIT NO. 12 294 19 PLAINTIFFS' EXHIBIT NO. 13 297 PLAINTIFFS' EXHIBIT NO. 14 301 20 PLAINTIFFS' EXHIBIT NO. 15 306 PLAINTIFFS' EXHIBIT NO. 16 317 21 PLAINTIFFS' EXHIBIT NO. 17 340 PLAINTIFFS' EXHIBIT NO. 18 341 22 PLAINTIFFS' EXHIBIT NO. 19 344 PLAINTIFFS' EXHIBIT NO. 20 373 23 PLAINTIFFS' EXHIBIT NO. 21 375 PLAINTIFFS' EXHIBIT NO. 22 379 Page 13 1 COMES GREGORY GEORGE ENAS, CALLED BY 2 THE PLAINTIFF, AND AFTER FIRST BEING DULY SWORN, 3 WAS DEPOSED AND TESTIFIED AS FOLLOWS: 4 DIRECT EXAMINATION 5 BY MR. SMITH: 6 Q. Would you state your name 7 please, sir? 8 A. Gregory George Enas. 9 Q. E-N-O-S? 10 A. E-N-A-S. 11 Q. How old a man are you, sir? 12 A. I'm thirty-seven years old. 13 Q. Where do you live? 14 A. Indianapolis. 15 Q. Address please. 16 A. XXXXXXXXXXXXXXXXXXXXXXX 17 XXXXXXXXXXXXXXXXXXXXX. 18 Q. Are you employed? 19 A. I'm employed at Eli Lilly. 20 Q. What's your job at Eli Lilly, 21 sir? 22 A. I'm the Director of Statistical 23 and Mathematical Sciences. 24 Q. Director of Statistical and Page 14 1 Mathematical Sciences? 2 A. Yes. 3 Q. What is your educational 4 background, starting with high school? 5 A. I graduated from high school in -- 6 from Modesto, California, Grace Davis High 7 School. 8 Q. When? 9 A. 1974, graduated from Biola 10 College in 1978, Los Angeles, California. 11 Q. You say that was in 1978? 12 A. Yes, degree in mathematical 13 sciences, B.S. I graduated in 1982 from the 14 Medical College of Virginia, Richmond, Virginia, 15 Ph.D. in biostatistics. 16 MR. RUIZ: What college was that? 17 THE WITNESS: Medical College of 18 Virginia. 19 Q. Is that a division of the 20 University of Virginia? 21 A. Virginia Commonwealth 22 University. 23 Q. And your degree there is Ph.D? 24 A. In biostatistics. Page 15 1 Q. Did you get a Masters or was 2 this degree that you got from the Medical College 3 of Virginia something that included Masters type 4 work? 5 A. I did not get a Masters degree. 6 I went straight for the doctorate. 7 Q. Any other educational 8 backgrounds? 9 A. No. 10 Q. Did you work while you were at 11 Biola College? 12 A. Yes, I did. 13 Q. What did you do? 14 A. I was a resident assistant my 15 last two years at Biola as a junior and senior 16 and I worked during the summers. One summer I 17 worked at a summer camp and another summer I 18 worked with youth, another summer I did 19 landscaping. 20 Q. Did you work while you were in 21 the Medical College of Virginia? 22 A. I had a graduate assistantship 23 and lectured in a class on statistics. 24 Q. After you got your degree, did Page 16 1 you begin working in the private sector? 2 A. Yes. 3 Q. Tell me about the job that you 4 first took after getting your Ph.D in 5 biostatistics. 6 A. I came to Eli Lilly and took a 7 job as a senior statistician. 8 Q. When -- do you remember your 9 specific start date at Lilly? 10 A. July 5, 1982. 11 Q. How long were you a senior 12 statistician? 13 A. Approximately five and a half 14 years. 15 Q. All right. What was your next 16 job at Lilly? 17 A. I became a research scientist 18 at Lilly. 19 Q. That would have been in 20 approximately '87, '88? 21 A. Yes, as of January 1, 1987. 22 Q. Was that a promotion to become 23 a research scientist? 24 A. Yes, it was. Page 17 1 Q. And how long did you hold that 2 job? 3 A. Approximately two and a half 4 years. 5 Q. And what was your next job? 6 A. In May, I believe of 1989, I 7 became a Manager in Statistical and Mathematical 8 Sciences. 9 Q. All right. And how long did 10 you hold that job? 11 A. Until approximately three 12 months ago when I -- so that would be 13 approximately three years -- I'm sorry, 14 approximately four years until May of or June of 15 this year. 16 Q. In June of 1993, did you assume 17 your present position as the Director of 18 Statistical and Mathematical Sciences at Eli 19 Lilly and Company? 20 A. Yes. 21 Q. What is a biostatistician as 22 opposed to a regular statistician? 23 A. A biostatistician is trained as 24 a regular statistician but the few departments in Page 18 1 the country that offer degrees in biostatistics 2 allow that person to engage in medical 3 consultation work throughout their graduate 4 studies. 5 MR. RUIZ: I'm sorry, did you say a 6 biostatistician is trained as a what? 7 THE WITNESS: Is trained as a regular 8 statistician. 9 Q. Does that mean that you took 10 courses in medical school? 11 A. The Department of Biostatistics 12 is housed within the School of Basic Sciences 13 within the Medical College of Virginia. 14 Q. But the Medical College of 15 Virginia turns out medical doctors, correct? 16 A. Yes. 17 Q. Probably nurses? 18 A. Yes. 19 Q. Maybe technicians such as x-ray 20 technicians and things of that nature? 21 A. I believe so. 22 Q. And sounds like they turned out 23 biostatisticians also? 24 A. Yes. Page 19 1 Q. My question is, did you take 2 courses that medical doctors would take, like 3 anatomy, for instance? 4 A. I did not take anatomy. Most 5 of my course work is in pure statistics. I had 6 an additional course in toxicology, for example, 7 that others outside of the school of 8 biostatistics might take as well. 9 Q. So there might have been 10 medical doctors in your toxicology class, is that 11 right? 12 A. In my particular class, I don't 13 believe there were any medical doctors in that 14 particular class. 15 Q. Medical students? 16 A. I don't believe there were any 17 medical students. I believe most of the students 18 in my particular toxicology class were students 19 in the School of Basic Sciences, those that were 20 working for Ph.Ds or Masters degrees in some 21 basic science. 22 Q. When you were in school at the 23 Medical College of Virginia, did you take any 24 courses in psychiatry? Page 20 1 A. No. 2 Q. Did you take any courses in 3 psychology? 4 A. No. 5 Q. Did you attend any hospital 6 rounds with teachers there in the Medical College 7 of Virginia? 8 A. No. 9 Q. Did you take any chemistry 10 courses at the Medical College of Virginia? 11 A. No. 12 Q. Did you take any biochemistry 13 courses at the Medical College of Virginia? 14 A. No. 15 Q. Did you take any pharmacology 16 courses at the Medical College of Virginia? 17 A. No. 18 Q. Did you take any 19 psychopharmacology courses at the Medical College 20 of Virginia? 21 A. No. 22 Q. Have you ever taken any courses 23 of that nature anywhere? 24 MR. MYERS: Of what nature? Page 21 1 MR. SMITH: Those courses I just 2 listed. 3 A. I've taken a chemistry course 4 at Biola College, one year of general chemistry. 5 Q. That's where you study atoms 6 like, things of that nature, basic compounds, 7 basic chemistry? 8 A. We studied molecules and atoms. 9 Q. Atoms make up molecules, right? 10 A. Yes. 11 MR. MYERS: We're really on our way 12 now. 13 Q. Did you take up -- I mean did 14 you take any psychology courses at Biola College? 15 A. I believe so. 16 Q. You took a basic psychology 17 course? 18 A. Introduction to psychology. 19 Q. Anything else other than that 20 introductory course? 21 A. No. 22 Q. Do you consider yourself expert 23 in psychology? 24 A. No. Page 22 1 Q. Do you consider yourself expert 2 in psychopharmacology? 3 A. No. 4 Q. Did you consider yourself 5 expert in pharmacology? 6 A. No. 7 Q. You are strictly a 8 biostatistician? 9 A. Yes. 10 Q. I guess the distinction I'm 11 trying to draw, Doctor Enas, is what's the 12 difference in a biostatistician and a straight 13 statistician, could it be said that you are a 14 statistician? 15 A. Yes. 16 Q. It looks like that you served 17 as a -- initially when you went with Lilly as a 18 senior statistician? 19 A. Yes, that is the title for 20 entry level Ph.D statisticians. 21 Q. Could you say you're also a 22 Ph.D statistician? 23 A. Yes, I am a Ph.D statistician. 24 Q. But you have a specialty in Page 23 1 biostatistics? 2 A. Yes, my degree is in 3 biostatistics, that is because the department I 4 graduated from was a statistics department but it 5 was housed within a medical school and the title 6 was biostatistics. 7 Q. But did you take any courses on 8 biochemistry or biology, zoology, anything of 9 that nature that makes you familiar with 10 physiology or how the human body works or reacts? 11 A. No, not at the Medical College 12 of Virginia. 13 Q. Anywhere? 14 A. I took a human genetics class 15 at Biola College. 16 Q. All right. 17 A. I took a one semester biology 18 course at Biola College, and that's it. 19 Q. Did you take courses in 20 statistics that -- at the Medical College of 21 Virginia that dealt with medical related issues? 22 A. Yes. 23 Q. Is that what makes you a 24 biostatistician or is it simply because you got Page 24 1 your statistics degree in a university that was 2 housed at a medical school? 3 A. Yes, and the curriculum. For 4 example, at the Medical College of Virginia in 5 the Department of Biostatistics it included 6 courses such as epidemiology which I did take. 7 Q. Any other courses specifically 8 designed in the medical field? 9 A. I've already mentioned the 10 toxicology course I took there. 11 Q. Anything else? 12 A. I took a course in survival 13 analysis. 14 Q. Survival analysis? 15 A. Yes. 16 Q. What's that? 17 A. It's statistical methodology 18 related to studies of disease where mortality or 19 other end points, where you measure time to a 20 certain event taking place. There are certain 21 statistical techniques that are appropriate for 22 those kinds of analysis and that course dealt 23 with those. 24 Q. I see, anything else? Page 25 1 A. Not that I can remember. 2 Q. Are there any other 3 subspecialties of Ph.D statisticians other than 4 biostatisticians that are recognized? 5 A. There are other names. 6 Q. Okay. 7 A. For these types, there are 8 biometricians. 9 Q. Biometricians? 10 A. Yes. 11 Q. What do they do? 12 A. They generally do what a 13 statistician does or biostatistician, it's just a 14 different name, different schools have different 15 departments with different names, that's the 16 other common name besides agrobiostatisticians or 17 statisticians. 18 Q. Any other subspecialties? 19 A. No. 20 Q. For instance, is there a 21 subspecialty in statistics that maybe studies 22 astronomy or oil and gas or the environment or 23 things of that nature? 24 A. Statistics can be applied Page 26 1 across many disciplines, I'm not familiar with 2 any programs that are specifically designed to 3 deal with oil and gas or astronomy, but the 4 practice of statistics and statistical thinking 5 can be applied across many disciplines. 6 Q. How would you define the 7 practice of statistics? 8 A. Statisticians are trained to 9 think objectively in terms of data and 10 variability and uncertainty. Statistical 11 thinking is the primary realm in which a 12 statistician deals and that's basically dealing 13 with uncertainties and variability and the 14 scientific method. 15 Q. Let me see if I can get that. 16 You said statisticians are trained to think 17 objectively in terms of variabilities? 18 A. Dealing with variability and 19 uncertainty in data and the use of the scientific 20 method. 21 Q. Let me see if I got this right. 22 Statisticians are trained to think objectively in 23 terms of variabilities and uncertainty in the use 24 of the scientific method? Page 27 1 A. And the use of the scientific 2 method. 3 Q. In analyzing a particular 4 problem, is that right? 5 A. That's one application, yes. 6 Q. What are some other 7 applications? 8 A. Designing a study to address a 9 question analyzing data, helping others in a 10 consulting fashion to look at their data and come 11 to reasonable conclusions based on their data. 12 Q. Is statistics a certain 13 science? 14 A. Yes. 15 Q. Is it basically a sophisticated 16 mathematical scientific process? 17 A. Mathematics is a tool that 18 statisticians use and sometimes the mathematics 19 can be sophisticated, other times it may not be, 20 it's not necessarily so. 21 Q. Can statistics be used to 22 determine the cause of a particular event? 23 A. Not by itself. 24 Q. Why? Page 28 1 A. It's only a tool. There -- by 2 itself in its mathematical form, it's only a tool 3 that one can use to look at certain aspects of a 4 situation. 5 Q. Once you come or once you use 6 this tool, can you come to a scientific certainty 7 concerning the cause of an event? 8 A. No. 9 Q. In other words, you can't tell 10 me as a statistician that if I flip a coin, the 11 probabilities are equal with respect to whether 12 or not that coin is going to come up heads or 13 tails, can you not? 14 A. Possibly. 15 Q. Are they equal? 16 A. If it's a fair coin, then I can 17 tell you, if it's not, then I may not be able to 18 tell you. 19 Q. Assuming that it's a fair coin? 20 A. Yes. 21 Q. It has heads on one side and 22 tails on the other side, probabilities are equal 23 that it will come up heads or tails, correct? 24 A. Yes. Page 29 1 Q. But you can't tell me at any 2 particular instance if I flip the coin once that 3 it's going to come up either heads or tails? 4 A. That's correct. 5 Q. All you can tell me from a 6 statistical standpoint in that analysis is that 7 there's an equal probability that it's either 8 going to come up heads or tails? 9 A. Yes. 10 Q. That's a pretty simple 11 statistical analysis, is it not? 12 A. Yes. 13 Q. It can become complicated 14 though, can it not? 15 A. Yes. 16 Q. The reason is because there's 17 variables, did you say? 18 A. The statistical method involves 19 variables and uncertainties in measuring lots of 20 things. 21 Q. Let me go back with you and 22 cover what you did during these different periods 23 of time before we get into esoteric questions 24 about statistics. When you were a senior Page 30 1 statistician at Eli Lilly, I believe you said you 2 began on July 5, 1982? 3 A. Yes. 4 Q. You did what, what was your 5 first job as senior statistician? 6 A. I worked within the statistics 7 department helping other statisticians designing 8 and analyzing clinical trials. 9 Q. Clinical trials? 10 A. Yes. 11 Q. And clinical trials with 12 respect to what particular drug? 13 A. I can't remember all the 14 particular drugs I worked on, there were a number 15 of them. 16 Q. Did that include Fluoxetine or 17 Prozac? 18 A. Not initially. 19 Q. At any time while you were a 20 senior statistician, did you work on clinical 21 trials on Fluoxetine or Prozac? 22 A. Yes. 23 Q. When do you recall beginning 24 first working on trials involving Fluoxetine? Page 31 1 A. It was a few years into my 2 career there, I can't remember the exact date or 3 time. 4 Q. '84, '85 maybe? 5 A. I can't remember. 6 Q. What was your first assignment 7 with respect to clinical trials on Fluoxetine? 8 A. I believe I was asked to take 9 over responsibilities for some clinical trials in 10 obesity and bulimia. 11 Q. What was your job going to be 12 in connection with those trials? 13 A. My job was to analyze the data 14 and help write the reports that would then go to 15 the regulatory agencies to earn approval for 16 those particular indications. 17 Q. Did you have to have any input 18 in the design of the clinical trial or were you 19 merely involved in analyzing data that was being 20 received and stored in the computers? 21 A. In some cases, I analyzed what 22 was there and in other cases I was involved as a 23 consulting statistician when the trials were 24 being designed. Page 32 1 Q. This would be in the protocol 2 stage? 3 A. Yes. 4 Q. And the people organizing the 5 clinical trial would come to you and ask you 6 questions like how many people do we need to use 7 in this clinical trial to determine a particular -- 8 or get a valid analysis of a particular 9 situation? 10 A. Yes. 11 Q. How do you make that decision? 12 A. One has to know exactly what 13 the question is that the investigator wants to 14 answer, that's the key. When one knows the 15 research question of interest, then the 16 investigator usually has a primary measurement 17 that they want to take in individuals to answer 18 that question and when that primary measurement 19 is known. Then one needs to know how similar 20 patients to those that would be enrolled in the 21 upcoming trial, how they responded or how they 22 were observed on that particular measurement, 23 because you need an estimate of how variable that 24 measurement is across the population, and taken Page 33 1 with that information -- and then the 2 investigator must provide some sense of what kind 3 of result do they want to see in this particular 4 trial, what's the current state of nature and 5 what do they want to see now, and given where you 6 are now and where you want to be, for example, 7 what kind of treatment effect do you want to see, 8 then the statistician can use that information to 9 calculate a sample size for that particular 10 trial. 11 Q. So, in determining the sample 12 size of a particular trial, the statistician has 13 to first focus on the particular question 14 involved? 15 A. Right, it has to be explicit. 16 Q. Beg your pardon? 17 A. It has to be a very explicit 18 quantifiable research question. 19 Q. And then he has to know the 20 result or the goal that the investigators are 21 attempting to achieve? 22 A. Yes, what would the 23 investigator wish to demonstrate, that would make 24 a clinically meaningful difference. Page 34 1 Q. So, if the investigator wanted 2 to support the hypothesis that a particular 3 medication increased blood pressure, you might 4 need a certain number of individuals, but if he 5 had the hypothesis -- wanted to have the 6 hypothesis that a certain medication didn't 7 affect blood pressure one way or the other, you 8 would need a different number of individuals, is 9 that what you're saying? 10 A. The number of individuals 11 depends on the research question. 12 Q. I understand it depends on the 13 question but does it depend on the goal that's 14 going to be achieved too? 15 A. Yes, using the scientific 16 method, you have a null hypothesis. 17 Q. Known hypothesis? 18 A. Null hypothesis. 19 Q. Null hypothesis? 20 A. That's right. 21 Q. N-U-L-L? 22 A. Yes, that describes the current 23 state of knowledge, and then you set up an 24 alternative hypothesis, this clinically Page 35 1 meaningful difference, what do I wish to 2 demonstrate with this new treatment for example 3 and then so you can compare your data when it 4 comes in against the null, if it doesn't fit with 5 the null hypothesis at the reasonable level, then 6 you can say I reject the null hypothesis and I 7 accept the alternative hypothesis and -- yes. 8 Q. Do you recall what some of the 9 initial questions with respect to Prozac were 10 posed to you in this phase while you were a 11 senior statistician? 12 A. I don't recall the exact 13 questions. 14 Q. You say they were in the 15 obesity and bulimia trials? 16 A. Yes. 17 Q. Approximately, what was the 18 exact question? 19 A. We wished to show the efficacy 20 and safety of Fluoxetine in patients with bulimia 21 and in patients with obesity. 22 Q. That was an ultimate question, 23 wasn't it? I mean that's the ultimate goal that 24 you wanted to do was to demonstrate that it was Page 36 1 efficacious and safe for treatment of obesity and 2 bulimia, correct? 3 A. Yes. 4 Q. I would assume though that 5 there was a step of questions that is going to 6 have to be analyzed to get to that conclusion, 7 correct? 8 A. Yes. 9 Q. My question to you is what were 10 the questions that you were in particularly -- in 11 particular addressing in those obesity and 12 bulimia trials? 13 A. I don't remember the exact 14 terminology, but for example in obesity, we 15 wanted to show some -- we wanted to show weight 16 loss in obese patients and so we would look for a 17 difference comparing the placebo and Fluoxetine 18 and look for an appreciable or clinical 19 meaningful difference of a certain number of 20 pounds weight loss. 21 Q. So you as a statistician would 22 attempt -- would you define what, as a 23 statistician what an adequate weight loss was? 24 A. No. Page 37 1 Q. Who would do that? 2 A. The medical experts define 3 what's a clinically meaningful difference. 4 Q. So they came to you and said 5 four pounds over a specific period of time is a 6 clinicallyly meaningful difference? 7 A. That's an example. 8 Q. They would then come -- then 9 what input would you have into that particular 10 formula? 11 A. Then, as I mentioned 12 previously, I would ask them, okay, you're 13 looking for an average of four pound difference, 14 what kind of variability do you see in these 15 types of patients, some lose less than four, some 16 lose more than four, show me the variability 17 there. And using that information, some of that 18 may be gathered from previous studies or 19 literature, wherever the clinician feels the most 20 pertinent information resides, and use that to 21 calculate sample size for the upcoming clinical 22 trial. 23 Q. Is all this work sort of built 24 on top of the other in this clinical trial Page 38 1 statistical analysis or design, you've always got 2 to take into consideration what's gone before? 3 A. That's the scientific method, 4 yes. 5 Q. So, how would you know that you 6 needed to study two hundred people to make -- to 7 achieve a goal of four pounds weight loss across 8 the statistical -- across the clinical trial? 9 A. Well, once you know the 10 treatment difference that you wish to detect and 11 the variability, then you set -- and these are 12 set for you many times, what your type one and 13 type two error rates are. In order to get a new 14 pharmaceutical approved in the United States in 15 the psychopharmacology division, a type one error 16 rate of .05 is generally the standard. That 17 means that there's a one in twenty chance that 18 you will claim there is a difference when there 19 really isn't and you want that to be small and 20 it's set at .05 generally. 21 Q. Okay. So you're saying that 22 just by chance one in twenty individuals, one of 23 those twenty individuals might have lost weight 24 regardless of medical treatment? Page 39 1 A. No, not individuals in studies 2 of a similar type. One, assume -- go back to the 3 null hypothesis, assume the null hypothesis is 4 true i.e., there is no difference say between the 5 placebo and another compound, you run twenty 6 trials, if the null hypothesis is true in one out 7 of twenty using the statistical methods, you're 8 going to claim there's a difference when there 9 really isn't, you want to set that to be very low 10 because you don't want to put treatments on -- 11 the FDA does not want to approve treatments that 12 are not efficacious. 13 Q. So the .05 is something that is 14 established by the Food and Drug Administration? 15 A. Generally, it's been 16 established even before the FDA got involved, 17 through tradition or whatever, statistical 18 tradition, but the FDA has taken that as a 19 standard. 20 Q. What -- is that because it's an 21 approved and recognized scientific standard in 22 the statistics area? 23 A. I'm not sure exactly but many 24 investigators for decades now have been using the Page 40 1 .05 level. 2 Q. Do you approve of that level as 3 an adequate level? 4 A. As a statistician, numerically 5 it's my opinion it is correct, but again we are 6 the servants of science so we have to yield to 7 the scientist to determine what level is correct 8 for their particular situation, but the FDA has 9 generally acknowledged the .05 level. 10 Q. Greg, you were explaining what 11 you were going to do and I interrupted you. 12 A. Well, that's part of this 13 equation, there's a formula for calculating 14 sample size for example and so you set -- your 15 type one, whenever you have this treatment 16 difference, you have the variability across the 17 population. And then another factor in the 18 sample size formula is what chance do you want 19 this trial to have if there is a true treatment 20 difference, if there is benefit, for example, 21 what chance do you want this trial to show that 22 difference. And you want that to be real high 23 because if you're going to do a study, the 24 investigators want to be reasonably sure that Page 41 1 there is benefit, they pick it up, so you set 2 that to be at eighty percent or ninety percent 3 and that's another factor in the equation. 4 Q. So, are you going to say that 5 eighty percent of the people are going to have to 6 gain four pounds of weight before you consider 7 that as a statistically meaningful trial? 8 A. No. 9 Q. Okay. 10 A. What I'm saying is that if I 11 ran a hundred trials of the same kind of trials, 12 one hundred of them, and indeed the treatment did 13 produce the weight loss that the clinician wanted 14 to see, if I ran a trial at eighty percent power, 15 eighty of those one hundred trials would show 16 that difference. 17 Q. All right. Are you saying that 18 statistically speaking the more trials you 19 conduct the more likely you are to have an 20 accurate result? 21 A. Not necessarily. 22 Q. Why? 23 A. What I'm saying is that if I 24 ran a hundred trials and there was a true Page 42 1 treatment benefit, eighty of those -- at each 2 trial there's variability here, eighty of those 3 trials would allow one to distinguish the 4 treatment difference, twenty of them would show 5 that there was no difference between treatments 6 even though there really was, so if by chance 7 this particular trial, there's a twenty percent 8 chance that it might be the one where no 9 treatment difference shows up but in actuality 10 there was a real treatment difference, so that's 11 where the variability comes into play. 12 Q. During that period of time, did 13 you do any other work other than work on obesity 14 and bulimia in the statistical area? 15 A. During which period of time? 16 Q. That period of time when you 17 were a senior statistician, I think you said it 18 was the initial two years I guess between July, 19 1982 and approximately mid 1984 or early '85? 20 A. Yes, I forget the exact times, 21 but yes, there were other projects that I worked 22 on. 23 Q. Did you work on depression 24 studies -- I want to limit myself right now to Page 43 1 Fluoxetine-Prozac studies, you told me about 2 obesity and bulimia, any other studies? 3 A. With respect to Fluoxetine, 4 those were the only studies I worked on at that 5 point in time. 6 Q. Were you working on studies 7 involving other drugs? 8 A. Yes, I was. 9 Q. Such as? 10 A. Such as Pergolide. 11 Q. Pergolide? 12 A. Yes. 13 Q. What's that? 14 A. It's a dopamine agonist for 15 Parkinsons patients. 16 Q. Anything else? 17 A. I helped a statistician do some 18 work on Penbutolol in the cardiovascular area. 19 Q. How about Tomoxatin, did you do 20 any work on Tomoxatin during that period of time? 21 A. I'm not sure if I worked on 22 Tomoxatin during that period of time. I have 23 done some work with Tomoxatin, I can't recall 24 specifics but I'm not sure if I worked on that Page 44 1 during that period of time. 2 Q. Tomoxatin was another 3 antidepressant being investigated by Lilly? 4 A. As far as I understand, it was 5 under investigation to determine whether it was 6 an antidepressant or not. 7 Q. Okay. Any other medications 8 that you worked with or compounds that you worked 9 with during that initial two, two and a half year 10 period of time? 11 A. I worked with Cinoxacin or 12 Cinobac which is registered for urinary tract 13 infections. 14 Q. Cinoxacin? 15 A. Yes. 16 Q. All right. 17 A. Nabalone, it's an antimedic for 18 use in cancer patients. 19 Q. Did you work with Doctor Stark 20 on that? 21 A. No. 22 Q. Doctor Paul Stark? 23 A. No. 24 Q. Did you know Doctor Paul Stark? Page 45 1 A. I met him, I never worked with 2 him. 3 Q. Any other medications? 4 A. Those are the primary ones. 5 Q. Who was your supervisor during 6 that period of time? 7 A. Charles Sampson. 8 Q. After that initial two years, 9 what did your job duties involve as a senior 10 statistician? 11 A. Well, what I mentioned to you 12 goes beyond the initial two years. 13 Q. All right. 14 A. It encompasses my tenure as a 15 senior statistician, I worked on a variety of 16 projects. 17 Q. Was all of this work on these 18 drugs done in connection with ongoing clinical 19 trials of these drugs? 20 A. Either ongoing or trials that 21 had been completed and I was asked to help 22 analyze data. 23 Q. Did your work involve anything 24 other than statistics as a result of clinical Page 46 1 trial data? 2 A. I contributed to the statistics 3 alone as a statistician, I analyzed data. 4 Q. What I mean but was it all 5 clinical trial data? 6 A. Are you -- please clarify. 7 Q. Well, you could have, I guess, 8 been a statistician involving and analyzing 9 market -- potential market work for new products 10 or balance sheets on Lilly over the last five 11 years or projections with respect to where a 12 particular drug -- what usage a particular drug 13 might have. 14 A. My work primarily involved 15 clinical trials, but I did some one-off things 16 with data that were not clinical trial data. 17 Q. You called that one-off? 18 A. Well different projects, just 19 different projects that were not clinical trial 20 related. 21 Q. But you used the term one-off? 22 A. One-off, in other words that's 23 not my primary focus but every so often a project 24 would come by that wasn't clinical trial related Page 47 1 that I would help out with. 2 Q. Any other work you did while a 3 senior statistician at Lilly during that initial 4 July, 1982, until January, 1987? 5 A. My work primarily revolved 6 around clinical trials. 7 Q. Can you give me any estimate of 8 the percentage of your work that involved the 9 Fluoxetine or Prozac trials during that period of 10 time? 11 A. During that time as a senior 12 statistician? 13 Q. Yes, sir. 14 A. Collectively, I would say just 15 approximately twenty percent, twenty-five 16 percent. 17 Q. Then in January of 1987, you 18 became a research scientist? 19 A. Yes. 20 Q. By the way, was Charles Sampson 21 your supervisor throughout the time you were a 22 senior statistician? 23 A. Yes. 24 Q. What was his title? Page 48 1 A. Initially it was manager and 2 then he was promoted to director, so he became 3 director, I'm not sure when that took place. 4 Q. I assume it's Doctor Sampson? 5 A. Yes. 6 Q. What's Doctor Sampson's 7 position now? 8 A. He is Director of Decision 9 Sciences. 10 Q. Director of Decision Sciences? 11 A. Yes. 12 Q. At Eli Lilly and Company? 13 A. Yes. 14 Q. In January of 1987, you were 15 given the title of research scientist, is that 16 correct? 17 A. Yes. 18 Q. Did your job duties change any? 19 A. No. 20 Q. Did you continue to work on the 21 Fluoxetine-Prozac clinical trials during that 22 period of time? 23 A. Yes. 24 Q. Was it exclusively obesity and Page 49 1 bulimia? 2 A. I can't say for sure whether it 3 was exclusively obesity and bulimia. 4 Q. Well, Prozac was approved for 5 marketing for treatment of depression in December 6 of 1987, as I recall. Is that correct, is that 7 your understanding? 8 A. I'm not sure. 9 Q. Did you do any work in 10 connection with securing Food and Drug approval 11 for treatment of Prozac in depression? 12 A. Not primarily, no. 13 Q. Then what were you doing, if 14 you weren't working on Food and Drug approval, 15 what were you doing in connection with the Prozac 16 clinical trials? 17 A. I was working on securing 18 approval for obesity and bulimia. 19 Q. The obesity and bulimia 20 clinical trials are still ongoing with respect to 21 Food and Drug Administration approval, correct? 22 A. I'm not sure if there are any 23 ongoing trials right now, the application is 24 still awaiting decision by -- Page 50 1 Q. It's not been formally acted on 2 one way or the other as far as you know? 3 A. Yes. 4 Q. But you don't think there are 5 any bulimia and obesity trials ongoing at this 6 time? 7 A. I'm not sure. 8 Q. You're not sure? 9 A. I'm not sure. 10 Q. You're not working on any of 11 them? 12 A. No. 13 Q. Up to January of 1987 when you 14 became a research assistant -- research scientist 15 at Lilly, did you ever do any statistical work in 16 connection with analyzing the issue of 17 suicidality in connection with the Fluoxetine 18 clinical trials? 19 A. No. 20 Q. Have you ever done any work in 21 connection with analyzing suicidality with 22 respect to Fluoxetine? 23 A. Yes. 24 Q. When did you first engage in Page 51 1 that job function? 2 A. I'm not sure of the exact date 3 or time. 4 Q. Would it have been after 5 January 1, 1987? 6 A. Yes. 7 Q. Would it have been before May 8 of 1989 when you became a Manager in the 9 Statistical and Mathematical Science Division? 10 A. I'm not sure, it was either 11 while I was a research scientist or while I was a 12 manager. 13 Q. And what work did you do in 14 that connection? 15 A. My basic function as a manager 16 was to be a sounding board for the group of 17 statisticians who were analyzing data relevant to 18 that question. So I would do what a manager 19 does, I'm there as a sounding board, give advice, 20 to provide resources and get tasks accomplished 21 and things like that. 22 Q. When approximately was that, 23 can you pin it down any? 24 A. Well, these roles were as Page 52 1 manager roles so I assume it was after I became a 2 manager. 3 Q. So it would have had to have 4 been before, I mean after January -- May, 1989? 5 A. It may have been, I'm not sure 6 because I was -- people who were working on these 7 issues may have been coming to me while I was 8 still a research scientist. But as a senior 9 person within the department, the senior people 10 are used as group leaders and people to be 11 sounding boards and give advice so I'm not sure 12 what time I started giving advice and being a 13 sounding board and getting resources, it doesn't 14 necessarily have to corroborate with being, 15 quote, a manager, I might have been just an 16 experienced person. 17 Q. What did you do, what was that 18 first assignment that you did in connection with 19 suicide and Fluoxetine? 20 A. It was reviewing analyses and 21 reports where the data analyses were presented. 22 Q. What analyses and reports did 23 you review? 24 A. I'm not sure exactly which Page 53 1 analyses and reports I reviewed but they all 2 dealt with the issue of suicidality. 3 Q. What were you examining, data? 4 A. I was examining reports which 5 included descriptions, verbage, tables with 6 numbers, reports that would ultimately go to the 7 Food and Drug Administration or other reviewers, 8 I would be involved with reviewing those reports. 9 Q. It sounds to me like you were 10 not directly involved but you were coming on to 11 oversee or -- 12 A. My primary responsibility was 13 to oversee the process. 14 Q. Okay. Who were you overseeing? 15 A. Some of the -- the 16 statisticians who were involved with actually 17 analyzing the data and working with the 18 clinicians and drafting these reports. 19 Q. Who were those statisticians? 20 A. Bruce Dornseif was the initial 21 one. 22 Q. Dornseif? 23 A. Yes, D-O-R-N-S-E-I-F. 24 Q. S-E-I-F? Page 54 1 A. Yes. 2 Q. All right. Was he senior or 3 junior to you? 4 MR. MYERS: In terms of age, rank, 5 experience? 6 MR. SMITH: Rank. 7 A. He was a senior statistician 8 who got promoted, I forget when, to research 9 scientist. 10 Q. Was he promoted before you? 11 A. No, he reported to me. 12 Q. Oh, he reported to you. 13 A. Yes. 14 Q. He was a Ph.D statistician 15 also? 16 A. Yes. 17 Q. Was he a biostatistician, do 18 you recall? 19 A. He got his Ph.D in 20 biostatistics. 21 Q. Where did he get his degree, do 22 you remember? 23 A. Medical College of Virginia. 24 Q. Did you recruit him? Page 55 1 A. Yes. 2 Q. Any other statisticians from 3 the Medical College of Virginia working at Lilly 4 other than you and Doctor Dornseif? 5 A. Doctor Dornseif does not work 6 at Lilly. 7 Q. I understand that, but 8 currently? 9 A. Oh, currently, yes, Sharon 10 Symanowski. 11 MR. MYERS: Is the question who are the 12 statisticians now or at the time this review or 13 analysis was going on? 14 MR. SMITH: At the time the analysis 15 was going on. 16 A. I'm sorry, I thought you said 17 now. No, that is it. Not Sharon Symanowski, she 18 is a new hire. 19 Q. Oh, you probably thought I was 20 looking for more -- 21 A. I thought you said who are the 22 biostatisticians from Virginia Medical College 23 working at Lilly now. 24 Q. Who were the other Page 56 1 statisticians that were reporting to you during 2 this suicide analysis in connection with Prozac? 3 A. Vin Rampey. 4 Q. Van? 5 A. Vin, Alvin. 6 Q. Alvin, spell the last name. 7 A. R-A-M-P-E-Y. 8 Q. Was he senior or junior to you? 9 A. He reported to me. 10 Q. Anybody else? 11 A. Doug Faries, F-A-R-I-E-S. 12 Q. He reported to you also? 13 A. Yes. 14 Q. Anybody else? 15 A. Mary Saylor, Susan Holman, Mike 16 Wilson, those are the statisticians primarily 17 involved with the analysis, day to day analysis. 18 Q. Are they all Ph.D 19 statisticians? 20 A. No. 21 Q. How about Miss Saylor? 22 A. No. 23 Q. Miss Holman? 24 A. No. Page 57 1 Q. And Mister Wilson? 2 A. No. 3 Q. Only -- well, how about Rampey? 4 A. Yes. 5 Q. Faries? 6 A. Yes. And Janet Bosomworth is 7 another, she's not a Ph.D statistician. 8 Q. Is she a statistician? 9 A. Yes. 10 MR. RUIZ: What was the last name, 11 please? 12 THE WITNESS: Bosomworth. 13 Q. Do you recall what the data was 14 that was being analyzed by the statisticians that 15 you were reviewing? 16 A. Not specifically. 17 Q. Do you recall generally? 18 A. Yes. 19 Q. What was it? 20 A. It was data reflecting 21 suicidality measures in patients from clinical 22 trials. 23 Q. So this was data only from 24 clinical trials? Page 58 1 A. Not necessarily. 2 Q. What other sources of data was 3 there that was being analyzed? 4 A. There were -- there are many 5 sources of data that, you know, for example, 6 spontaneous reports, post-marketing reports. 7 Q. What else? 8 A. The clinical trial data, that's 9 primarily the source. 10 Q. Was there any other data that 11 was obtained in that review, in that analysis? 12 In other words, you've got to define what data 13 you're looking at before you can make an 14 intelligent comment concerning any representation 15 reflected by that data, don't you? 16 A. The primary analyses were 17 performed on the clinical trial data. 18 Q. But you're saying that it's 19 your recollection that the spontaneous reports 20 from post-marketing experience was also included 21 in this data? 22 A. People were looking at all the 23 data. 24 Q. Okay. Were they including it Page 59 1 in that data that they were reporting that was 2 being analyzed? 3 MR. MYERS: When you say included, do 4 you mean together with the clinical trial data? 5 MR. SMITH: Yes. 6 A. No those data were kept 7 separate from clinical trial data. 8 Q. But the statistical analysis 9 that you and these individuals were performing in 10 connection with suicidality and Fluoxetine, did 11 that involve only analysis of the clinical trial 12 data? 13 A. Primarily the clinical trial 14 data, I don't remember any other analysis beyond 15 the clinical trial data. 16 Q. Well, would that be significant 17 to know in connection with whether or not 18 Fluoxetine had any part in suicidality, to know 19 what the various sources of the data were? 20 A. It's important to know all the 21 various sources of data but the clinical trial 22 data became the primary data base because it 23 gives a valid, unbiased answer because you have 24 controlled groups concurrently treated with Page 60 1 treatment groups, the spontaneous reports do not 2 provide adequate control and hence your 3 inferences or conclusions aren't -- 4 Q. Why do you say it was unbiased 5 data, the clinical trial data was unbiased? 6 A. I mean that you are comparing 7 patients who are treated with placebo who were 8 randomized to that group versus patients who were 9 randomized to say Fluoxetine or another 10 tricyclic, and hence you have the same kinds of 11 patients taking placebo as you do taking 12 Fluoxetine, the only difference in those patients 13 is that half of them say got Fluoxetine, the 14 other half got placebo. That's the randomized 15 controlled trial which has become more or less a 16 gold standard in medical science because it 17 provides clean cut answers with respect to the 18 treatment effect because all the other patient 19 characteristics wash out because of the 20 randomization process. 21 Q. Do you know anything about how 22 the randomization process was achieved in the 23 clinical trials done on Fluoxetine? 24 A. Please clarify that. Page 61 1 Q. Do you know anything about the 2 randomization process that occurred in the 3 clinical trials with respect to Prozac? 4 A. I do know that the patients 5 were randomized in these various studies to 6 either treatment group or placebo, that's all I 7 know about the randomization. 8 Q. Well, if there were errors in 9 that randomization, that might affect the outcome 10 of the data from a statistical standpoint, 11 couldn't it? 12 A. Errors in the randomization 13 could affect your analysis. 14 Q. Do you know whether there were 15 any errors in the randomization in the Fluoxetine 16 clinical trials? 17 A. It's my opinion that there were 18 no errors in the randomization process. 19 Q. In all of your statistical 20 analyses, did you ever make any inquiries to 21 determine whether or not the randomization 22 process was done correctly? 23 A. There are a number of things 24 that the statisticians do to look at that, for Page 62 1 example we compare the patients at entry between 2 the treatment groups. For example, if you get 3 all the males in one group and all the females in 4 another group, that might raise a flag there that 5 something happened, so we look at basic 6 demographic and physiological characteristics and 7 compare them at baseline or before randomization, 8 before the treatment is administered to see if 9 there is basically an equal spread. 10 Q. My question to you is do you or 11 any of your colleagues make an investigation into 12 whether or not there had been any errors reported 13 in connection with the randomization process or 14 the reporting of the randomization in connection 15 with the Fluoxetine trials? 16 A. I don't believe any 17 statisticians went to the actual randomization 18 and investigated that, all we did was look at how 19 it turned out with respect to the patient 20 characteristics. 21 Q. What do you mean you looked at 22 how it turned out with respect to the patient 23 characteristics? 24 A. For example, were the genders Page 63 1 approximately equal across the treatment groups. 2 Q. Did you check anything other 3 than gender? 4 A. A number of variables typically 5 are -- 6 Q. What was checked? 7 A. I can't remember specifically 8 all that was checked, but basic demographic 9 statistics, the outcome, the primary outcomes are 10 assessed, like the Ham-D instrument, is that 11 basically equal at baseline. 12 Q. But if there were problems in 13 randomization in connection with the Fluoxetine 14 clinical trial, you didn't know about it, did 15 you? 16 A. If there were problems, no. 17 The data looked reasonable at baseline, that is 18 what a statistician does, we -- the statistics 19 group does not go back to the actual hands on 20 randomization process, i.e. how the physician 21 actually gets the treatment and dispenses it to 22 the patient. 23 Q. Did you make any determination 24 in connection with whether or not the Page 64 1 randomization process in assigning a particular 2 Almedica kit to particular patients and 3 particular Almedica numbers to particular 4 patients was accurate or inaccurate? 5 A. I don't know what the 6 statisticians did with respect to Almedica and 7 the patient numbers. 8 Q. You didn't yourself do anything 9 to confirm that that was done properly? 10 A. No. 11 Q. Or that the reports were done 12 properly? 13 A. No. 14 MR. MYERS: What reports? 15 MR. SMITH: The reports in connection 16 with the randomization. 17 A. What reports? 18 Q. Didn't somebody provide a check 19 to make sure that there was some to ensure that 20 randomization was done accurately and properly? 21 A. There is a group who was 22 involved with labeling of bottles using Almedica, 23 the clinical trials materials group. 24 Q. Did you check with the clinical Page 65 1 trials materials group? 2 A. I did not personally. 3 Q. Why do you say that the data 4 from the clinical trial process was unbiased? 5 A. That's a statistical term. 6 Q. Unbiased is a statistical term? 7 A. Yes, it's a statistical term, 8 it means that when you use randomization, as you 9 conduct your experiment, you're putting patients 10 of similar characteristics on one treatment and 11 matching them to similar characteristics on the 12 other treatment, so what you come up with at the 13 end is a clear picture of the treatment effect 14 because, for example, you hopefully get equal 15 numbers of males and females in each group and so 16 those possible confounding variables that might 17 also explain treatment differences are not 18 existent, you're left with a clear treatment -- 19 indication of treatment effect and that's 20 statistically what we mean by unbiased. There 21 are no other factors that are biasing that 22 treatment difference that you're observing. 23 Q. But whether or not a physician 24 is biased in his reporting with respect to a Page 66 1 particular increase or decrease in a Ham-D score, 2 that's not what you're speaking of? 3 A. Not at all. I'm referring to 4 the act of randomization and coming up with only 5 one factor being able to explain the treatment 6 difference and that is the treatment itself 7 because all the other baseline factors have been 8 spread out across the treatment groups. 9 Q. Why is that different from the 10 result from the data on the spontaneous reports 11 on suicidality? 12 A. Spontaneous reports come from a 13 variety of patients and the question you have to 14 ask yourself is compared to what. Those patients 15 might be a certain type of patient, there may be 16 no untreated group that you can compare it to to 17 see if there is a difference or not. That's why 18 the randomized control trial is taken as the gold 19 standard often because you can get a clear 20 picture. 21 Q. Well, if a person was -- if a 22 person committed suicide while on the same dosage 23 of Fluoxetine and was the same age and the same 24 sex, say one person was in clinical trials and Page 67 1 the other person wasn't, are you saying that that 2 individual that's in the clinical trial is going 3 to be of more statistical importance than the 4 individual who was reported as a spontaneous 5 report? 6 A. No. 7 Q. All right. 8 A. What I am saying is that those 9 two patients, even though on age and sex they 10 might be the same, they might be very different 11 patients with respect to some other biological or 12 physiological mechanism. 13 Q. Can you compare the spontaneous 14 reports with respect to suicide attempts on 15 Fluoxetine with other antidepressant medications? 16 A. You may not be able to. 17 Q. Why? 18 A. Again the patients may be 19 different, you don't -- you may not have a 20 comparable set of patients. That's why again we 21 run randomized control trials and that's why to 22 prove efficacy you have to inquire randomized 23 control trials in general because you have a 24 comparison which leaves no other explanatory Page 68 1 variable except the treatment itself, no other 2 confounding factor. 3 Q. So is it your opinion that from 4 the statistical standpoint, that if there are no 5 other confounding factors then the treatment has 6 to be the effect, is that what you're saying? 7 A. Yes, in the randomized control 8 trial. 9 Q. So if a suicide occurred during 10 the randomized control trials where there was no 11 other explanation for an individual's suicide 12 other than the fact that they were consuming 13 Fluoxetine, then that was to be statistically 14 significant in whether or not that individual 15 committed suicide as a result of ingesting 16 Fluoxetine? 17 A. No. 18 Q. All right. Where am I wrong? 19 A. Because you have to compare 20 Fluoxetine to placebo, it's not that just looking 21 at Fluoxetine patients that were treated with 22 Fluoxetine and committed suicide, you have to 23 look at the rates in both groups and if the 24 placebo rate and the Fluoxetine rate are similar, Page 69 1 then there may be no reason to believe that it's 2 the drug that's inducing any suicidality because 3 we have seen the same rate in those that weren't 4 treated with Fluoxetine and the same rate in 5 those that were treated with Fluoxetine. 6 Q. You can't use statistics to 7 determine whether or not a particular drug has a 8 causative effect -- 9 A. That's right, it goes back to 10 our -- 11 Q. -- with respect to a particular 12 condition? 13 A. That's right. 14 MR. MYERS: Let him finish his question 15 before you start talking, otherwise you're going 16 to drive our court reporter mad. 17 THE WITNESS: I'm sorry. 18 MR. MYERS: I'm sorry, were you 19 finished? 20 MR. SMITH: I didn't mean to interrupt 21 you. Can you give me that last 22 answer so I'll make sure that I understand it and 23 that we've got it? 24 (THE COURT REPORTER READ BACK THE Page 70 1 REQUESTED TESTIMONY.) 2 A. I'm saying that in the 3 randomized trial, you compare like patients with 4 like patients and if you see a difference and the 5 only explanatory variable was that one group got 6 a drug and one group got a placebo, then you 7 attribute that difference to the treatment itself 8 and nothing else. 9 Q. Okay. But so I understand it, 10 you cannot prove or disprove from statistics 11 whether or not Fluoxetine causes suicidality in a 12 particular depressed person, can you? 13 A. Statistics can help you look at 14 the aggregate evidence, and in my opinion, 15 there's no evidence from the aggregate that 16 Fluoxetine does induce suicidality. So 17 statistics again deals with populations, it deals 18 with samples. In one individual patient you 19 cannot use statistics, per se, as we've been 20 discussing here, but in the aggregate that's 21 where statistical thinking helps you weigh that 22 evidence. 23 Q. Even in the aggregate, you 24 can't prove or disprove that Fluoxetine causes Page 71 1 suicidality with respect to ingestion of Prozac, 2 can you? 3 A. Statistics can help you or 4 allows you to come up with a conclusion as to 5 whether or not Fluoxetine induces suicidality or 6 not, so it does allow you to come up with a 7 conclusion that you feel strongly about because 8 the data supports that. 9 Q. All you can say from a 10 statistical standpoint is whether or not it's 11 more likely or less likely, can't you? 12 A. That's right, it weighs the 13 evidence for you and so when you look at the 14 aggregate evidence, statistics helps you bring 15 that evidence together and decide one way or the 16 other. 17 Q. But the statistical aspect of 18 it is only one piece of evidence, isn't it, 19 Doctor? 20 A. That's correct. 21 Q. In other words, it's only a 22 piece of the scientific question, isn't it? 23 A. Yes. 24 Q. And it's not the -- in all Page 72 1 deference to you as a statistician, it's not the 2 be all or end all, is it? 3 A. No. 4 Q. And you wouldn't presuppose to 5 give testimony in any of these cases that any of 6 our clients who had committed suicide after 7 ingesting Prozac did not commit suicide as a 8 result of their ingestion of Prozac, would you? 9 MR. MYERS: Before he answers, let me 10 just object to the form, because if the focus of 11 your question is an individual patient or 12 plaintiff, that approach is a medical opinion and 13 I object to the form, but if he can answer it, go 14 ahead. 15 A. I was going to tell you that I 16 am not a physician and hence as a statistician, I 17 can only deal with the samples and populations. 18 I cannot, because I'm not a physician, deal with 19 individual patients. 20 Q. From a statistical standpoint, 21 you cannot produce statistical evidence 22 concerning whether or not Fluoxetine played a 23 part in any particular individual's suicide, can 24 you? Page 73 1 A. That's correct. 2 Q. And you don't intend to offer 3 statistical evidence in any particular case 4 concerning whether or not a particular individual 5 statistically did or did not commit suicide as a 6 result of ingestion of Fluoxetine, are you? 7 MR. MYERS: Hold on, if the question is 8 directed to him, he can answer. When you say 9 you, I'm assuming you mean him and not Lilly. 10 MR. SMITH: Yes. 11 A. Here's what I can offer as a 12 statistician, I can help -- 13 Q. What I want you to do is answer 14 the question that was directed to you. 15 MR. MYERS: You answer it as best you 16 can, Greg. 17 A. As a statistician, I can help 18 the physicians and others involved look at the 19 data and determine in this body of evidence that 20 we have here from the controlled clinical trials, 21 does there seem to be a preponderance of suicides 22 in one group or the other or do they seem to be 23 comparable. That's what I, as a statistician, 24 can offer with respect to that particular data Page 74 1 base. 2 Q. But that's not conclusive at 3 all with respect to whether or not a particular 4 individual did or did not commit suicide as a 5 result of the ingestion of Fluoxetine, is it? 6 A. That's not a statistical 7 question and hence I cannot answer that. 8 Q. Statistics involves logic, does 9 it not? 10 A. Yes. 11 Q. From the logistic -- from a 12 logical standpoint, whether or not a group is 13 more likely or less likely to commit suicide 14 doesn't have any bearing on whether or not a 15 particular individual did or did not commit 16 suicide as a result of ingestion of Fluoxetine? 17 MR. MYERS: Well, before he answers, 18 let me object to the form and your description of 19 what is or is not logical because I don't think 20 that's at all relevant as to what your 21 description of what's logical is. But if he can 22 answer the question based on his knowledge and 23 his expertise, go ahead. 24 MR. SMITH: Quit coaching him, Larry. Page 75 1 MR. MYERS: I'm not coaching. 2 MR. SMITH: If you continue to coach 3 him, I'm going to -- 4 MR. MYERS: Yes. 5 MR. SMITH: -- lose my cool. 6 MR. MYERS: All right. 7 MR. SMITH: I might even point some 8 fingers. 9 MR. MYERS: I've developed a resistance 10 to that. 11 MR. SMITH: Please read back my last 12 question. 13 (THE COURT REPORTER READ BACK THE 14 REQUESTED TESTIMONY.) 15 Q. Does it? 16 A. That's correct. The data that 17 we have though after weighing the evidence, we 18 can then take that and say for other patients, we 19 would see statistics of sampling, taking samples 20 and then generalizing to broad populations, we 21 can say that for other patients given the 22 evidence we have now, there does not seem to be 23 an increased risk of suicide given the data that 24 we have. So for any particular patient in this Page 76 1 big population that may be treated, for example, 2 we can say as statisticians, the evidence does 3 not support any increased risk with that 4 treatment, so for those particular patients in 5 the future that might be taking this drug, that's 6 what the scientific method is all about, 7 extrapolating from the sample and the data you 8 have to the general population. 9 Q. So you're saying even though it 10 doesn't present a risk to the general population 11 from a statistical standpoint? 12 A. Yes, from a scientific -- 13 Q. But from a scientific 14 standpoint, you will acknowledge that whether or 15 not it poses a risk to a particular individual 16 cannot be statistically determined, correct? 17 A. Not entirely. 18 Q. Not entirely statistically able 19 to be determined or not entirely correct? 20 A. Not entirely, because 21 statistics again is only dealing with the data 22 that we have at hand and then generalizing to a 23 broader population, it takes the individual 24 patients that you have data on and then makes Page 77 1 conclusions based on that data that are 2 generalizable to the broad population. But it's 3 a medical question that I'm not inclined to 4 answer with respect because it goes into 5 mechanisms and things like that that I cannot 6 answer at all. I can only take the sample that 7 we have of data and take that data and 8 extrapolate beyond that. 9 Q. And your extrapolations can 10 only be as good as the data that is given you, 11 correct? 12 A. Yes. 13 Q. If the data given you is flawed 14 in any respect, that's going to affect your 15 extrapolations and your conclusions from the 16 statistical standpoint? 17 A. The extrapolations are only as 18 good as the data that you're building your case 19 upon. 20 Q. And the more error there is in 21 that underlying data that you formed statistical 22 analysis of, the more likely it is that your 23 statistical analysis may result in conclusions 24 that are erroneous? Page 78 1 A. What do you mean by more error 2 in the data? 3 Q. The more error in the 4 underlying data, if the underlying data is 5 flawed, then it's statistically likely that your 6 statistical conclusions may be flawed? 7 A. Yes, if the data is wrong, then 8 your inferences, your conclusions will be wrong. 9 But that's more than just statistics, that's 10 logic again, that's a general. 11 Q. Obviously as a statistician, 12 one of your goals is to -- is an attempt to get 13 as clean a set of data as possible, correct? 14 A. Yes. 15 Q. And you are doing -- have done 16 as well as you can to ensure that your data is 17 accurate, correct? 18 A. Yes. 19 Q. But if it's inaccurate, it can 20 report -- result in reporting statistically 21 inaccurate information? 22 A. Possibly. 23 Q. Let me talk with you a little 24 bit about some basic statistics. In statistics, Page 79 1 in biostatistics, or in any type of statistics, 2 if you're trying to find a particular subgroup of 3 individuals, is there a particular analysis that 4 you have to make, in other words in order to get 5 to that subgroup of individuals? 6 A. Are you saying I have a data 7 base and I'm looking for different subgroups 8 within the existing data base that I have? 9 Q. Right. 10 A. Yes, there are techniques that 11 help one look for subgroups. 12 Q. How many people would you have 13 to look to to determine whether or not a 14 particular subgroup exists within a particular 15 population? 16 A. Well, that all depends on what 17 trait or characteristic of that subgroup you're 18 looking for. 19 Q. Let's say you were looking for 20 a subgroup of people who became suicidal while 21 taking Fluoxetine. 22 A. Okay. 23 Q. How many people would you have 24 to examine to determine whether or not in Page 80 1 particular there existed a subgroup of 2 individuals who were not suicidal before taking 3 Fluoxetine but after taking Fluoxetine became 4 suicidal? 5 A. I don't know how many 6 individuals that would take. 7 Q. Why? 8 A. It depends on the underlying 9 characteristics of that population, what 10 treatments were administered, how you're 11 measuring suicidality, is it a very precise 12 measurement or is it you know, full of lots of 13 things, it depends on those things and once you 14 iron those out, you are more in a position to 15 determine actual number of patients. 16 Q. Do you think you could ever 17 find that subgroup from a purely statistical 18 standpoint? 19 A. No. 20 Q. Why? 21 A. Because it's much more than 22 statistical. 23 Q. All right. 24 A. You are talking Page 81 1 reproducibility, for example, just because you 2 see something in one data base, you want to look 3 at other evidence to confirm that effect. That's 4 why, for example, at the FDA they desire more 5 than one clinical trial to get a drug approved 6 because they want to see reproducibility of 7 effect. So statistics again is a tool but it's 8 not the end all to doing science, it's a tool of 9 science. 10 MR. SMITH: Please read back that 11 answer. 12 (THE COURT REPORTER READ BACK THE 13 REQUESTED TESTIMONY.) 14 Q. Are you speaking of 15 rechallenge? 16 A. No. 17 Q. Is rechallenge a valid analysis 18 from a statistical standpoint? 19 A. Statistically -- the scientific 20 process in statistics can be applied to any type 21 of study. I cannot speak to rechallenge because 22 that's more of a medical or pharmacological 23 question, but statistics and design, experimental 24 design, can be applied to any type of clinical Page 82 1 trial. 2 Q. If you have produced a result 3 on more than one occasion under the exact 4 circumstances, would that tend to lead to the 5 conclusion that the condition was a result of the 6 particular circumstance, it was produced on more 7 than one occasion? 8 A. That's right, if those trials 9 corroborate one another, see the same trend 10 across multiple studies, yes, that corroboration 11 builds a case. 12 Q. Or you rechallenge. You know 13 what I mean by rechallenge, do you not? 14 A. Clarify. 15 Q. Where you expose an individual 16 again to something that has produced a result on 17 one occasion and in that you reexpose them to 18 that same condition under the same circumstances 19 repeatedly and you get the same results from a 20 statistical standpoint and from a logical 21 standpoint that is more evidence that a result 22 might have occurred from a particular cause, is 23 that correct? 24 A. Yes, more from a logical Page 83 1 standpoint. 2 Q. Okay. 3 A. Not so much from a statistical 4 and rerandomized sense because, for example, time 5 is elapsing, the patient might be changing. So 6 just because you repeatedly challenge them, there 7 might be other factors, but from a randomized 8 controlled trial standpoint, the statistics can 9 help you design that type of trial. 10 Q. But logically rechallenge is a 11 valid -- 12 A. I'm not a pharmacologist. 13 Q. I'm not talking about 14 pharmacology, Doctor, I'm talking about logic. 15 A. That is -- rechallenge is one 16 kind of study I have heard of that people use in 17 many experimental situations to demonstrate, 18 right, repeated effects which build a case that 19 something is going on. 20 Q. Well, that's done in clinical 21 trials, is it not? 22 A. I have heard of it being done 23 in clinical trials, I'm not familiar with -- 24 Q. It's certainly done in Page 84 1 medicine? 2 A. Yes, from what I can tell. 3 Q. You go into your allergist and 4 they expose you to several different allergic 5 substances -- 6 A. Right. 7 Q. That's sort of a rechallenge, 8 isn't it, correct? 9 MR. MYERS: Well, I object to the form 10 only to the extent you just described one 11 challenge with a substance and then you're 12 talking about rechallenge so -- 13 A. I would say that's not 14 rechallenge from my perspective because they got 15 different substances each time. To me a 16 rechallenge is you take the same substance 17 repeatedly. 18 Q. For instance, a particular 19 pollen, if you give that individual a particular 20 pollen and they start sneezing each time, that 21 makes the particular pollen more suspect, does it 22 not? 23 A. The same pollen over time, yes, 24 there might be other factors that can explain Page 85 1 that, for example, the patient is getting older, 2 but that's one possible cause. 3 Q. Is it statistically possible to 4 determine whether or not a subgroup exists who 5 are not suicidal before taking Fluoxetine but 6 became suicidal after taking Fluoxetine? 7 A. In certain situations 8 statistics can help one make that determination. 9 Q. All right. Have you ever done 10 that while you were employed by Eli Lilly and 11 Company, made any attempt to determine whether or 12 not a subgroup exists or doesn't exist in which 13 individuals were not suicidal prior to ingestion 14 of Fluoxetine and became suicidal thereafter? 15 A. From what I can understand, 16 that's what much of our data base is, patients 17 who were not suicidal, then were randomized 18 either placebo or Fluoxetine and some cases 19 tricyclic, and then statistics can help 20 physicians and other investigators compare those 21 treatments to see if there's any increase or 22 decrease in those rates. 23 Q. Did you ever find, as a result 24 of those studies -- or did you ever conclude that Page 86 1 there was not a subgroup that existed that were 2 becoming suicidal because of Prozac? 3 A. No, we did not find that. We 4 found that overall, over the entire data base, 5 the picture was no different between the 6 treatment, the Fluoxetine was not inducing 7 suicidality. 8 Q. But you didn't find that there 9 did not exist a subgroup which were of 10 individuals which were becoming suicidal because 11 they were being exposed to Fluoxetine? 12 MR. MYERS: Excuse me, Paul, this group 13 did or did not exist, is that the question? 14 Q. You never established that the 15 group did not exist that became susceptible to 16 becoming suicidal when exposed to Fluoxetine? 17 A. May I use the restroom? That's 18 a -- could you write that out for me? 19 MR. SMITH: That may be the most 20 variable -- 21 MR. MYERS: May we take a short break? 22 MR. SMITH: Sure. 23 (A SHORT RECESS WAS TAKEN.) 24 MR. MYERS: Would you read back the Page 87 1 last question. 2 (THE COURT REPORTER READ BACK THE 3 REQUESTED TESTIMONY.) 4 A. We have not seen any subgroup 5 in the data base where there is susceptibility in 6 Fluoxetine. 7 Q. Well, you knew or it was 8 obvious from the data presented to you that some 9 individuals committed suicide while taking 10 Fluoxetine, correct? 11 A. Yes. 12 Q. So there were a group of 13 individuals during the clinical trials that 14 committed suicide while receiving Fluoxetine 15 therapy? 16 A. Yes. 17 Q. So there is a group of 18 individuals that committed suicide while taking 19 Prozac, correct? 20 A. Yes. 21 Q. Your statistical analysis 22 simply compared that group, those people taking 23 Fluoxetine with other people during the clinical 24 trials that committed suicide, correct? Page 88 1 A. It compares Fluoxetine treated 2 patients with patients who were not taking 3 Fluoxetine. 4 A. That would be the placebo 5 group? 6 A. Placebo group. 7 Q. And you found that according to 8 your analysis in the data you reviewed that they 9 were not -- there was not a greater number of 10 instances of individuals who committed suicide in 11 a placebo group than in the Fluoxetine group? 12 A. We found there's no treatment 13 difference between the placebo group and the 14 Fluoxetine group. 15 Q. Let's don't talk about 16 treatment difference, let's talk about people who 17 committed suicide. 18 MR. MYERS: Are you talking about 19 numbers now? 20 MR. SMITH: Yes. 21 MR. MYERS: Committed suicides? 22 MR. SMITH: Right. 23 Q. The data you reviewed showed 24 people that committed suicides while taking Page 89 1 Fluoxetine during the clinical trials, correct? 2 A. Yes. 3 Q. And the data you reviewed 4 showed people who committed suicide who are on 5 the placebo, correct? 6 A. Yes. 7 Q. Your data also showed that 8 there were people who committed suicide during 9 the clinical trials who were on the comparitor 10 drug, correct? 11 A. Yes. 12 Q. Imiprimine or whatever the 13 other antidepressant was, correct? 14 A. Yes. 15 Q. Your statistical analysis only 16 established that there weren't larger numbers in 17 the Fluoxetine group who committed suicide than 18 those numbers in the placebo group and those 19 numbers in the comparitor group, correct? 20 A. It established that, yes, those 21 groups had similar rates, there weren't increases 22 or decreases. 23 Q. But with respect to whether or 24 not any of those individuals in the Fluoxetine Page 90 1 group that committed suicide -- you don't know 2 whether or not, from a statistical standpoint, 3 any of those individuals committed suicide as a 4 result of the ingestion of Prozac, do you? 5 A. That's correct. In the same 6 vein, we don't know from a statistical standpoint 7 whether those who received placebo committed 8 suicide because of placebo. We compared the 9 treatment groups and compared the rates. 10 Q. All you got was a comparison? 11 A. Well, that's the gold standard 12 from a randomized clinical trial is a comparison, 13 because you need to answer the question compared 14 to what and the randomized trial gives you that 15 comparison. 16 Q. But as far as whether or not 17 Fluoxetine caused a particular Fluoxetine patient 18 to commit suicide, you don't have any 19 determination of that, do you? 20 A. No. 21 Q. Now, as a biostatistician, are 22 you familiar with the fact that some people are 23 allergic to Penicillin? 24 A. I've heard of people who take Page 91 1 Penicillin that have reactions, but whether it's 2 allergic or not, I can't determine that to be an 3 allergic response or not. 4 Q. Are you familiar with the 5 prevailing scientific medical view that some 6 people have an adverse reaction to Penicillin 7 that can be quite serious? 8 A. I've heard that. 9 Q. You probably have been to your 10 doctor complaining of a particular condition such 11 as pneumonia, influenza, or something of that 12 nature and your doctor has asked you are you 13 allergic to Penicillin, correct? 14 A. Yes. 15 Q. Are you familiar with the fact 16 that it is a relatively minor amount -- number of 17 the general population who are allergic to 18 Penicillin? 19 A. No, I'm not familiar with that, 20 I don't know what the rate is. 21 Q. I understand that you don't 22 know what the rate is but do you understand that 23 it's a relatively minor number of individuals? 24 A. I've never thought about that, Page 92 1 I don't know if it's a minor or a lot of 2 individuals that have allergic reaction to 3 Penicillin. 4 Q. Well, you know that Penicillin 5 is a widely prescribed medication, do you not? 6 A. Yes. 7 Q. And you know that there are 8 less people who get Penicillin that get allergic 9 reactions to Penicillin, correct? 10 A. I would hope so. 11 Q. And that's because the majority 12 of patients don't get an allergic reaction to 13 Penicillin, correct? 14 A. Again, I don't have knowledge 15 of the actual rates. It seems logical that, yes, 16 that's true, but I haven't seen the evidence and 17 data. 18 Q. It wouldn't surprise you to see 19 statistics that indicated that the actual number 20 of individuals who are allergic to Penicillin is 21 a relatively small number of individuals, would 22 it? 23 A. Yes, but again that involves a 24 question compared to what, when you say Page 93 1 relatively small, that means you're comparing it 2 to some number. 3 Q. Everybody in the world, 4 correct? 5 A. Yes. 6 Q. Can you do that from a 7 statistical standpoint, compare -- make 8 comparisons of a phenomenon based on the entire 9 world population? 10 A. Numerically you can, but 11 whether -- 12 Q. Numerically you cannot? 13 A. Numerically you can. I said 14 you can compare one number to another number 15 numerically but you have to know if you are 16 comparing apples and apples or apples and 17 oranges, that's why you come back to the question 18 compared to what. Those two numbers may 19 represent two completely different things and the 20 actual numerical comparison may not be correct, 21 that's why we again go back to the randomized 22 trial because it provides an apples to apples 23 comparison. 24 Q. If the randomized trials are Page 94 1 indeed randomized -- 2 A. Yes. 3 Q. -- and if the data is 4 accurately reported -- 5 A. Yes. 6 Q. -- and if the data is based on 7 sound scientific principles? 8 A. Yes. 9 Q. Your statistical evidence can 10 only be as good as the underlying data? 11 A. Yes. 12 Q. But the issue of whether or not 13 a particular individual has committed suicide as 14 a result of taking Fluoxetine is a medical issue, 15 not a statistical issue, isn't it? 16 A. For that particular patient, 17 it's not a statistical issue, what statistics can 18 give you though is the likelihood that that's due 19 to a particular treatment or not and that's why a 20 randomized controlled data base tells us that the 21 likelihood is not there that's due to Fluoxetine, 22 we've seen similar rates in all the treatment 23 groups so the likelihood of any particular 24 individual based on that randomized data base is Page 95 1 the same. 2 Q. Isn't the key there likelihood 3 as opposed to whether or not it actually -- it 4 did or did not occur? 5 A. Right, statistics deals with 6 likelihoods. 7 Q. Statistics deals with 8 likelihoods, not certainties, correct? 9 A. Statistics deals with 10 likelihoods and likelihoods can range from very 11 uncertain to very certain. 12 Q. All right. But all you can say 13 is that from that data that you reviewed with 14 respect to whether or not Fluoxetine caused 15 suicidality, it was less likely of the group that 16 were analyzed, correct, that Fluoxetine caused 17 suicidality than any other treatment group? 18 A. We can say that it is unlikely 19 because -- 20 Q. It's unlikely? 21 A. Yes. 22 Q. Where does less likely become 23 unlikely? 24 A. Just that again we do not find Page 96 1 any significant difference between the treatment 2 groups using that randomized control data base. 3 Q. My question was, where does 4 unlikely -- where does less likely become 5 unlikely in a statistical analysis, is there a 6 particular percentage point? 7 A. No, again the .05 significance 8 level is the one that is generally agreed upon as 9 a standard, and once we run a urinalysis and you 10 compare your data and your resulting P values to 11 that .05 significance level, and if it doesn't 12 attain that significance level then you do not 13 reject the null hypothesis, so it's very 14 stringent. 15 Q. All right. Did you find that 16 in your review of the -- your statistical review 17 of the Fluoxetine and clinical trial data that 18 Fluoxetine was more likely to improve suicidal 19 thinking than other anti-depressants? 20 A. I believe that on the 21 comparison of one of the measures of suicidal 22 thinking on the Ham-D, which is the Ham-D item 23 three, they were demonstrated to be improvements 24 on Fluoxetine compared to placebo. Page 97 1 Q. What about comparitor drugs? 2 A. Compared to the tricyclics? 3 Q. Yes. 4 A. I'm not sure what the 5 significance level was there. 6 Q. What's your recollection 7 generally as to whether or not there was any 8 difference in the significance level between the 9 Fluoxetine treated group and the comparitor 10 treated group? 11 A. Well, that information is in 12 the, for example, the British Medical Journal 13 publication, I can't recollect the numbers right 14 now but I believe that information is in there. 15 Q. What's your recollection of 16 what that information is? You just said that you 17 had a recollection with respect to placebo. 18 A. Definitely, that's one I 19 remember the most. 20 Q. Why? 21 A. Because there you have patients 22 who are untreated with patients that are treated 23 with a particular agent, so that's one comparison 24 that's made. Page 98 1 Q. Well, isn't it significant as 2 to whether or not they're getting better on 3 Fluoxetine or another antidepressant? 4 A. Yes, it is. 5 Q. And so what was the 6 significance between Fluoxetine and the other 7 antidepressant? 8 A. To the best of my recollection, 9 I don't think there was any significant 10 difference between the two. 11 Q. All right. So Fluoxetine was 12 not more -- any more effective in reducing 13 suicidal Ham-D 3 scores than was other 14 antidepressants to which it was compared? 15 A. There was a greater difference 16 between Fluoxetine and placebo than Fluoxetine 17 and the comparitor drugs, I can't recall whether 18 there was a significant difference between 19 tricyclics. 20 Q. I'm talking about comparitors -- 21 Fluoxetine and comparitor drugs. 22 A. I don't know whether it's 23 significant at the .05 level or not, I would have 24 to look at the British Medical Journal to Page 99 1 determine that. 2 Q. Was this work that you were 3 doing a review of the data for the Beasley 4 British Medical Journal article? 5 A. That was part of my review work 6 was looking at that publication. 7 Q. Was that the majority of it? 8 A. That was a large part of it. I 9 was also involved with reviewing reports that 10 went, for example, to the Food and Drug 11 Administration, so publications were part of it 12 as well as regulatory reports. 13 Q. What other publications did you 14 work on other than the British Medical Journal? 15 MR. MYERS: On this issue? 16 MR. SMITH: Yes. 17 A. I have reviewed a number of 18 publications. 19 Q. What other publications? 20 A. Would you like titles or -- 21 Q. Title or author or both if you 22 can give them to me or publication even. I just 23 want the most information I can get from you. 24 A. I reviewed a number of Page 100 1 publications of which my staff were authors on 2 but I can't remember the titles, the authors 3 would be those that the names of my staff that 4 I've given you before, various ones with 5 co-author, various publications, I can't remember 6 the names of those publications. 7 Q. That would be Dornseif? 8 A. Yes, he's been a co-author on 9 some. 10 Q. Rampey? 11 A. Yes. 12 Q. And Faries? 13 A. Yes. 14 Q. How about Bosomworth? 15 A. Yes. 16 Q. I haven't seen any publications 17 by Mike Wilson, Mary Saylor or Susan Holman? 18 A. No, I don't believe -- well, 19 Mary Saylor, I believe, was an author on one. 20 Q. Which one? 21 A. The British Medical Journal 22 publication. 23 Q. The Beasley medical analysis? 24 A. Yes. But not Wilson or Holman. Page 101 1 Q. Any other publications you can 2 think of that you reviewed the statistical 3 standpoint? 4 A. With respect to suicidality? 5 Q. Yes. 6 A. I reviewed -- I can't give you 7 exact number of publications, but I reviewed 8 cohort publications as well as regulatory 9 reports. 10 Q. What? 11 A. Regulatory reports. 12 Q. That would be the safety 13 update? 14 MR. MYERS: Which one? 15 MR. SMITH: Any one of them. 16 MR. MYERS: Okay, a safety update. 17 A. I have been involved with a 18 number of regulatory submissions, the safety 19 update may or may not have been included in my 20 actual purview but I was involved as being a 21 reviewer. 22 Q. How about that submission to 23 the Food and Drug Administration when the Food 24 and Drug Administration was considering changing Page 102 1 the labels, having the special hearings 2 post-approval, did you participate in that 3 publication? 4 A. I reviewed those reports that 5 went down to the Food and Drug Administration. 6 Q. Let me see if I have this clear 7 and please bear with me, Doctor, I'm not trying 8 to be picky but I'm trying to understand what 9 you're saying. Can you base any conclusion on 10 whether or not Prozac causes a particular 11 depressed individual to become suicidal who is 12 not suicidal prior to taking Prozac based on a 13 statistical analysis only? 14 A. I cannot for a particular 15 individual -- 16 Q. All right. 17 A. -- using statistical analysis. 18 Q. Can you form any conclusion 19 based on statistics in any particular trial as to 20 whether a particular individual might become 21 suicidal -- might have become suicidal while on 22 Prozac? 23 A. Again -- 24 Q. Examining a particular clinical Page 103 1 trial? 2 A. Again in statistics, we haven't 3 examined the mechanism by which the suicides 4 happen in a particular individual, all we've done 5 as we've explained before is just compared the 6 sample populations. 7 Q. So the answer as I posed it is 8 no or the question as I posed it is no? 9 A. Yes, that's correct. 10 Q. Can you form any conclusion 11 based on a statistical analysis of whether in a 12 particular group of clinical trials that Prozac 13 causes suicidality in a particular subset of 14 individuals who become suicidal while on 15 Fluoxetine when you're examining a particular 16 group of clinical trials? 17 A. We have not found any such 18 subset of patients that have a particular 19 propensity for suicidal ideation while on 20 Fluoxetine. 21 Q. But you can't come to a 22 conclusion based on a statistical analysis of 23 whether in a particular group of clinical trials 24 that Prozac caused a particular individual within Page 104 1 that particular clinical trial to become 2 suicidal, can you? 3 A. We can conclude that given the 4 data we've looked at we've not found any subset 5 where that is true, where there is a propensity 6 for suicidal ideation on Fluoxetine. So given 7 the data we have, we can conclude as to this 8 point in time we haven't seen any evidence of 9 that. 10 Q. Well, there is a subgroup as 11 per the contraindications that Prozac should not 12 be given to individuals who are hypersensitive to 13 Prozac, isn't there? 14 MR. MYERS: Let me object to the form, 15 Paul. When you say a subgroup, do you mean 16 within the clinical trials? 17 MR. SMITH: Yes. 18 A. I'm not sure what evidence that 19 labeling is based upon, I'm not sure of that 20 evidence. 21 Q. In your review of the 22 Fluoxetine trials, didn't you find that there was 23 a subgroup that were hypersensitive to Prozac? 24 A. What do you mean by Page 105 1 hypersensitive? 2 Q. As that term is defined 3 medically. 4 MR. MYERS: He's not a doctor so I 5 object to the form. 6 MS. ZETTLER: It's a term that's used 7 by Lilly. 8 MR. MYERS: I understand, but he's not 9 a doctor. 10 THE WITNESS: I'm not sure. 11 MS. ZETTLER: I know he's not a doctor, 12 Larry. I'm saying it's a term defined by Lilly, 13 they have determined that there is a group of 14 people who are hypersensitive to Prozac and 15 Paul's question is how do they come to that 16 determination. 17 A. And again, my answer is I'm not 18 sure, I don't know how they define that subgroup 19 or what data that wording is based upon, I'm not 20 sure. 21 Q. You can't give any medical 22 conclusions, can you? 23 A. No, I'm a statistician. 24 Q. Don't intend to, do you? Page 106 1 A. That's correct. 2 Q. And the issue with respect to 3 whether or not a particular individual has become 4 suicidal while on Fluoxetine is a medical issue, 5 isn't it? 6 A. It's not a statistical issue. 7 I'm sure it's a medical issue, a pharmacological 8 issue, it may not just be purely medical, there's 9 lots of things that might contribute to that, 10 statistics in the randomized trial helps you 11 merit out those but it's not -- you have to rely 12 on the purely life sciences to help you determine 13 a particular patient. 14 Q. You didn't talk to any 15 individual who became suicidal during the 16 clinical trials, did you? 17 A. No. 18 Q. You didn't talk to any 19 physicians who reported that their patients 20 became suicidal while on the clinical trials, did 21 you? 22 A. No. 23 Q. You didn't talk to any family 24 members of individuals who committed suicide Page 107 1 while on the clinical trials either, did you? 2 A. No. 3 Q. And that's going to be good 4 evidence that will have a strong bearing on 5 whether or not a particular individual committed 6 suicide as a result of taking Fluoxetine, won't 7 it? 8 MR. MYERS: Let me object to the form, 9 good evidence. I'm objecting to the form, Paul. 10 MR. SMITH: He gave an answer and she 11 can't even hear what his answer is because of 12 your objection. 13 MR. MYERS: That's intentional, I want 14 to object before he answers but he can answer. I 15 object to the form and the use of the term good 16 evidence as being possibly a legal term, but he 17 can answer if he can from his knowledge and 18 understanding. 19 A. I said no. 20 Q. You said no, it's not good 21 evidence? 22 A. Yes, in my opinion. 23 Q. What would be good evidence in 24 your opinion? Page 108 1 A. From a purely statistical 2 standpoint, good evidence comes from randomized 3 trials where you're comparing patients who did 4 receive the agent in question versus those that 5 did not receive the agent, a placebo group, and 6 comparing those treatment groups to see if there 7 is any greater likelihood for patients treated 8 with Fluoxetine to have suicidality -- to have 9 suicidality. 10 Q. I understand -- 11 A. From my perspective, that is 12 good evidence. 13 Q. That's good statistical 14 evidence. 15 A. That's right, good statistical 16 evidence. 17 Q. I'm talking about evidence 18 concerning whether or not a particular individual 19 committed suicide as a result of taking 20 Fluoxetine, good evidence with respect to that 21 individual would be discussions with the 22 physician who is treating that patient, correct? 23 A. I can't address that because 24 I'm not a physician, I don't know exactly what Page 109 1 paradigm individuals use outside of statistics to 2 determine whether it's related or not. 3 Q. Do you have an opinion 4 concerning whether or not Fluoxetine has an 5 affect on the biochemical nature of individuals' 6 brains who ingest Fluoxetine? 7 MR. MYERS: Before he answers, I will 8 object to the form. That is a medical and 9 scientific opinion question that's outside of his 10 described area of expertise. If he has some 11 knowledge, he can answer. 12 MS. ZETTLER: Ask him as a research 13 scientist. 14 Q. Answer my question first. 15 A. As a statistician, the data I 16 have seen supports the fact that Fluoxetine is 17 safe and effective. 18 Q. No, that's not what I put to 19 you. 20 A. Okay. I'm telling you as a 21 statistician that's all I can say from the data. 22 Q. Do this for me, Doctor, would 23 you, listen to my question, respond to the 24 question as I ask it if you can. If you can't Page 110 1 respond to it as I ask it, then tell me. 2 A. Okay. 3 Q. And here's my question: Do you 4 have an opinion concerning whether or not 5 Fluoxetine has an affect on the brain chemistry 6 of individuals who consumed Fluoxetine? 7 MR. MYERS: Same objection. If you can 8 answer it, answer it, if you can't, tell him 9 that. 10 A. I do not have a medical 11 opinion. 12 Q. Do you have an opinion as the 13 Director of Statistics and Mathematical Sciences 14 at Eli Lilly and Company as to whether or not 15 Fluoxetine has an effect on the brain chemistry 16 of individuals who consumed Fluoxetine? 17 MR. MYERS: Same objection. 18 A. I have an opinion. 19 Q. What is that opinion, sir? 20 MR. CLEMENTI: If he has an opinion, 21 I'm going to join the objection to the question 22 previously asked and I'll join Larry's objection. 23 MR. RUIZ: We will have as well. 24 MR. CLEMENTI: Go ahead. Do you want Page 111 1 her to read the question back? 2 MR. MYERS: Do you remember the 3 question? 4 THE WITNESS: Does Fluoxetine have an 5 effect on the brain. 6 MR. MYERS: The brain chemistry, I 7 think Mr. Smith said. 8 MR. CLEMENTI: And that's the question 9 I'm putting my objection to. 10 MR. MYERS: If you can answer it, 11 answer it. 12 A. My opinion is I don't truly 13 know as a pharmacologist or physician what kind 14 of effect there is on the brain chemistry. Based 15 on the data and the numbers that I've seen, there 16 is some kind of effect that Fluoxetine has. 17 Q. All right. And you're 18 expressing that opinion in your current position 19 as the Director of Statistics and Mathematical 20 Science at Eli Lilly and Company, correct? 21 A. Yes. I don't have a medical 22 opinion as to whether it affects brain chemistry 23 or not, I've seen symposium and posters, but just 24 as a statistician looking at the numbers, it does Page 112 1 have an effect and that's why I said my opinion 2 is that it's safe and efficacious because I 3 haven't seen an effect of Fluoxetine working on 4 depression and positively helping patients with 5 depression, and so I'm attributing that effect 6 because they are in randomized studies again to 7 Fluoxetine. 8 Q. Well we would object to the 9 answer as being non-responsive totally to the 10 question that was put to you. Remember, you're 11 going to have the opportunity to answer any 12 questions, Doctor Enas, that your attorney would 13 like to put to you. Right now I'm asking you the 14 question, so you've got to give me the answers to 15 the questions I ask. 16 MR. MYERS: Don't pay any particular 17 attention to that. Ask him a question, Paul, 18 don't instruct him. 19 MR. SMITH: I'm trying to assist him. 20 I'm sure he's not giving me these non-responsive 21 answers purposely or as a result of any 22 instructions you may have given him, Mr. Myers. 23 I'm simply trying to assist him in making it 24 easier for us to get through this question and Page 113 1 answer. 2 MR. MYERS: Try another question. 3 MR. SMITH: And I wouldn't think he 4 would have intentionally given me an unresponsive 5 answer. 6 MR. MYERS: I can't imagine that he 7 would have, so try another question and see where 8 we go. 9 MR. SMITH: Don't direct me, Mr. Myers, 10 all right, on how to conduct this examination. 11 MR. MYERS: If you don't want to ask 12 another question, that would be fine as well. 13 Q. Do you have an opinion as 14 Director of Statistical and Mathematical Sciences 15 at Eli Lilly, Doctor Enas, concerning whether or 16 not Fluoxetine has an effect on individuals' 17 behavior? 18 MR. MYERS: Same objection. 19 A. Before I answer, I'd like to 20 understand the dialogue that's going on here 21 because -- 22 MR. MYERS: I object because you're not 23 a physician and that's a medical question. If 24 you know or if you have an opinion, tell him. If Page 114 1 you don't know or you don't have an opinion, tell 2 him that. 3 MS. LAWS: We do have the agreement 4 that one objection is good for all? 5 MS. ZETTLER: That's across the board. 6 MR. SMITH: Or bad for all. 7 A. I have stated my opinion and 8 that is based on the data I see. 9 Q. Is that it does have an effect? 10 A. There are differences between 11 placebo patients and Fluoxetine patients and I 12 can only attribute that to Fluoxetine and I gave 13 an example where you see improvement in 14 depression and I attribute that to Fluoxetine. 15 So it logically, in my opinion, not medical, but 16 in my opinion just as one who has run randomized 17 studies, that that effect is due to the 18 treatment. But how it works mechanistically, 19 that's a medical and pharmacologic opinion and I 20 cannot speak to that. 21 Q. And again my question was 22 directed to you as the Director of Statistical 23 and Mathematical Science at Lilly. I simply 24 asked you as Director of Statistical and Page 115 1 Mathematical Science, do you have an opinion 2 simply whether or not Fluoxetine has an effect on 3 individuals? 4 A. Yes. 5 Q. And what is that opinion, that 6 it does? 7 A. My opinion is that it does have 8 an effect. 9 Q. From a statistical standpoint, 10 could you say that if you had an adverse event in 11 only one instance, versus a non-adverse event in 12 a hundred instances, that that adverse event is 13 not related to the drug? 14 MR. MYERS: Let me object to the form. 15 When you say non-adverse event, you mean no 16 event? 17 MR. SMITH: Right. 18 A. So could you clarify a bit more 19 for me, you have no events in a hundred? 20 Q. Yes, one person who has -- who 21 reports increased sweating -- 22 A. Okay. 23 Q. -- as a result of -- while on 24 the drug. Page 116 1 A. Right. 2 Q. And he's only one in a group of 3 a hundred individuals who are given the drug, 4 correct, you follow me so far? 5 A. One out of a hundred got the 6 increased sweating. 7 Q. Right. Does that mean from a 8 statistical standpoint that that one individual 9 did not get sweating as a result of the drug? 10 A. No. 11 Q. All right. If you had one 12 individual out of a thousand who were given the 13 drug, does that mean that one individual did not 14 get sweating as a result of the drug? 15 A. That one individual did get 16 sweating? 17 Q. Right. 18 A. One out of a thousand, then you 19 said did not get sweating? 20 Q. One in a thousand reported 21 sweating. Does that mean that the sweating is 22 not a result of the drug because it only occurred 23 in one in a thousand times? 24 A. No. Page 117 1 Q. One in ten thousand, one 2 individual reports sweating out of ten thousand 3 people exposed to the drug, does that mean that 4 that one person did not get sweating as a result 5 of the drug? 6 A. That one person did get 7 sweating. 8 Q. Right. 9 A. I want to clarify because you 10 said did not get sweating. 11 Q. It's late in the day. 12 MR. MYERS: Statistically speaking, 13 you're still asking him. 14 A. As I understand your question, 15 you're saying did -- one individual out of ten 16 thousand got sweating, does that mean that the 17 drug caused the sweating. 18 Q. Right. 19 A. Statistically speaking, no. 20 Q. Does it mean it doesn't cause 21 the sweating? 22 A. No. Again you have to go back 23 to compared to what, one in a thousand compared 24 to what. Page 118 1 Q. What about one in a million? 2 A. No. 3 MR. MYERS: When you say no, your 4 answer is the same? 5 A. Yes, the same as the others. 6 Q. What about one in ten million, 7 what about one in four point five million, sounds 8 like a nice figure to me. 9 A. One out of four point five 10 million compared to what? Again, without a 11 comparison statistically -- 12 Q. One compared to the four point 13 five million receiving the drug, can you say from 14 a statistical standpoint that that one person 15 doesn't get sweating because of the drug? 16 A. Again, that person did get 17 sweating. 18 Q. Yes. And we're determining 19 causation to it, did the drug cause that one 20 person to get sweating from a statistical 21 standpoint? 22 A. No, again you cannot determine 23 that. 24 Q. Are you familiar with Page 119 1 statistics that indicate that depressed 2 individuals are more likely to commit suicide 3 than non-depressed individuals? 4 A. I'm not familiar with the 5 statistics themselves, I've heard that statement 6 before but I'm not familiar with the statistics. 7 Q. You've never seen any 8 statistical analysis concerning whether or not 9 that statement is accurate? 10 A. I may have but I can't recall 11 any numbers comparing depressed to nondepressed. 12 Q. Do you think it's an accurate 13 statistical analysis to say that depressed 14 individuals are more likely to commit suicide 15 than nondepressed individuals? 16 A. I can't comment on the accuracy 17 of that particular statement because I don't know 18 what data went into generating that statement, 19 I'd have to see the data and what mechanism by 20 which that data were generated to tell you 21 whether I believe that conclusion was valid or 22 not. 23 Q. Would you have to see 24 controlled clinical trials on that since that's Page 120 1 the gold standard? 2 A. Yes, that's the best way to 3 answer a question is using the randomized 4 controlled trial. 5 Q. In other words, before you 6 would accept the hypothesis that depressed 7 individuals are more likely to commit suicide 8 than nondepressed individuals, you would want to 9 see controlled clinical trials to support that 10 hypothesis? 11 A. It may not be doable in that 12 case but the randomized controlled trial is the 13 gold standard, you try to achieve that level if 14 possible. 15 Q. Do I understand that you're not 16 going to agree with me from the statistical 17 standpoint as the Director of Mathematics and 18 Statistics at Eli Lilly and Company that 19 depressed individuals are more likely to commit 20 suicide than nondepressed individuals? 21 A. From what I have read, I agree 22 with that statement. I just haven't seen the 23 actual -- I can't recall or have seen the actual 24 statistics that went into that statement. Page 121 1 Q. So are you telling me that as a 2 statistician you don't know whether or not those 3 statistics are accurate? 4 A. I have not seen the data that 5 led to those statements so I cannot testify to 6 the validity one hundred percent of that 7 statement, I need to see the data and the 8 mechanism, et cetera. 9 Q. And that data that you would 10 need to see would be controlled clinical trials? 11 A. It wouldn't need to be 12 controlled clinical trials because I'll take a 13 look at any kind of data. I'm saying that the 14 randomized controlled study, if we have it, is 15 the best by which we can make that determination. 16 Q. All right. If it weren't a 17 randomized controlled clinical study that made 18 that hypothesis, would you be less certain of 19 that hypothesis as being accurate statistically? 20 A. Most likely, yes. 21 Q. All right. 22 MR. CLEMENTI: Can we take a quick 23 break? 24 MR. SMITH: Sure. Page 122 1 (A SHORT RECESS WAS TAKEN.) 2 Q. Doctor, can you say from a 3 statistical standpoint that if a person becomes 4 less depressed that they are less likely to 5 commit suicide? 6 A. I cannot say that right now 7 based on the data. 8 Q. For the same reasons that you 9 were giving me earlier? 10 A. That's right, in an individual 11 patient. 12 Q. Did you find in your 13 statistical analysis of Fluoxetine in any respect 14 that from a statistical standpoint Fluoxetine 15 produced adverse reactions? 16 A. Based on the data, there are 17 adverse reactions that have higher rates in 18 Fluoxetine treated patients than placebo treated 19 patients. 20 Q. So from that standpoint, 21 Fluoxetine does have a causative adverse event 22 over and above that of the placebo nonmedicated 23 patients? 24 A. Just to clarify, statistics Page 123 1 can't prove causality -- 2 Q. All right. 3 A. -- but in the randomized 4 setting, it then leads more credence to causality 5 when you see a difference. 6 Q. Right. So it's more likely 7 that Fluoxetine caused that particular adverse 8 event from a statistical standpoint? 9 A. Yes. 10 Q. Correct? 11 A. Well, it's more likely that 12 those events will occur on Fluoxetine treated 13 patients than placebo treated patients. Again to 14 say it's causal, statistics doesn't deal with 15 those mechanisms by which that is caused. 16 Q. Either good or bad? 17 A. That's right. 18 Q. And those mechanisms have to do 19 with medical determinations concerning causality? 20 A. Yes. Statistics is one piece 21 of the puzzle but there's other pieces as well 22 that go outside of statistics. 23 Q. Would it be accurate to state 24 from a statistics standpoint that the more data Page 124 1 that is available, the more likely that you'll 2 get a more sophisticated and more comprehensive 3 statistical analysis? 4 A. You used a lot of terms there 5 that I don't know if my definition would equal 6 your definition, like comprehensive. I will 7 state that the more data you have, the more 8 confident one becomes in the conclusions. 9 MR. SMITH: All right. That's all I 10 have right now. 11 * * * * * * * * * * 12 CROSS EXAMINATION 13 BY MS. ZETTLER: 14 Q. Who told you that the FDA 15 recognizes the P value of .05 as -- I can't 16 remember the word you used, authoritative or -- 17 A. A significant level. 18 Q. A significant level. 19 A. There are FDA guidelines that 20 have been published. 21 Q. Have you actually read those 22 guidelines? 23 A. I have but I don't remember 24 explicitly whether they state .05 or not in the Page 125 1 guidelines but I have read those guidelines. 2 Q. And somebody at Lilly told you 3 that the FDA had stated that .05 is significant? 4 A. When I first joined Lilly, 5 someone told me that and then I've heard it from 6 lots of sources and seen proof of that by the 7 fact that they use that level in making 8 determinations. 9 Q. They use the level on their own 10 or did you say accept that level when it's used 11 by a drug company? 12 A. They use that level because any 13 reviewer can use whatever level they want to, the 14 statistician produces a P value and any reviewer 15 uses their own level to see if that's below or 16 above that level. 17 Q. Have you ever had any 18 conversations or communications with FDA 19 personnel? 20 A. Yes, I have. 21 Q. Okay. And were those -- any of 22 those conversations or communications related 23 directly to Fluoxetine? 24 A. Yes. Page 126 1 Q. On how many occasions did such 2 communications take place? 3 A. I can't recall. 4 Q. Do you remember any of those 5 conversations specifically or any of the 6 communications specifically? 7 A. I have been involved in 8 meetings where I may have spoken up but I was 9 primarily just an observer where other people 10 were talking back and forth and I know of at 11 least one meeting that I was involved with the 12 FDA, but beyond that I can't recall exactly what 13 communications I've had with FDA. 14 Q. Can you remember generally what 15 types of communications you would have with FDA 16 personnel? 17 A. In general or with respect to 18 Fluoxetine? 19 Q. With respect to Fluoxetine. 20 A. I was involved -- I was a 21 participant in one meeting we had dealing with 22 suicidality and the FDA. 23 Q. Where was that meeting held? 24 A. In the Washington, D.C. area, I Page 127 1 forget the meeting room but it may have been at 2 the FDA headquarters there in Rockville. 3 Q. When did that meeting take 4 place? 5 A. It was prior -- I believe it 6 was prior to the advisory hearing that was held 7 with respect to Fluoxetine and suicidality, I 8 can't remember the specific date. 9 Q. Do you remember what the 10 subject matter of the meeting was? 11 A. Fluoxetine and suicidality, 12 evidence for or against that. 13 Q. Was there more than one meeting 14 that you participated in with FDA personnel 15 regarding Fluoxetine and suicidality? 16 A. I can only remember one right 17 now where I was actually present but I'm sure -- 18 Q. You're sure there were more? 19 A. I'm unsure, right. 20 Q. Could this meeting have 21 anything to do with rechallenge protocol or 22 rechallenge clinical trials? 23 A. I believe I've heard that term 24 come up in a meeting, whether it was that Page 128 1 particular meeting or not, I'm not sure, but I've 2 heard that term come up in meetings with FDA. 3 (PLAINTIFFS' EXHIBIT NO. 1 WAS 4 MARKED FOR IDENTIFICATION AND 5 RECEIVED IN EVIDENCE.) 6 Q. Why don't you take a look at 7 the entire exhibit. Okay. Have you had a chance 8 to review Exhibit Number 1, Doctor? 9 A. Yes, not in depth but I have a 10 feel for what this is about. 11 Q. Okay. If I ask some questions 12 and you feel like you need to review it, go ahead 13 and let me know that, okay? 14 A. Yes. 15 Q. Does this refresh your 16 recollection as to whether or not you attended 17 the meeting discussed in this exhibit? 18 A. Definitely. 19 Q. Is this the meeting that you 20 recall we were just talking about a couple of 21 minutes ago? 22 A. Yes. 23 Q. Does it refresh your 24 recollection as to whether or not a rechallenge Page 129 1 protocol was discussed at that meeting? 2 A. Yes. 3 Q. Tell me when you first heard 4 about a rechallenge protocol? 5 A. I cannot tell you specifically 6 when I first heard about a rechallenge protocol. 7 Q. Can you give me an idea of how 8 long before this meeting occurred that you first 9 heard about it, was it a couple of months or a 10 year? 11 A. I cannot recall. 12 Q. Tell me what your understanding 13 of a rechallenge protocol as reflected in Exhibit 14 1 was. 15 A. This first exhibit does not 16 tell me what the rechallenge study, per se, was. 17 Like it addresses some points assuming that the 18 audience knows what the rechallenge study was and 19 then it says collect data to compare patients 20 enrolled in this particular protocol. It 21 doesn't, in my opinion, address the actual 22 rechallenge protocol itself as far as what it 23 was. 24 Q. Do you have a recollection of Page 130 1 what the actual rechallenge protocol was? 2 A. To the best of my knowledge, 3 there is a process here, as mentioned here, 4 there's this M-S-S-I-R in clinical trials. There 5 were some modifications that one physician here 6 who's a professor, Ivan Miller, was involved with 7 to come up with a valid or a usable instrument to 8 measure suicidality. That was the first step to 9 address what measurement tool we were going to 10 use to actually assess this phenomena. And then 11 secondly, once that is assessed then incorporate 12 that in a protocol whereby patients who had been 13 treated with antidepressants, not just 14 Fluoxetine, but any anti -- some particular types 15 of antidepressants would then be randomized to 16 see -- Fluoxetine and some other comparitor or 17 placebo arms, I'm not sure exactly how many arms, 18 but there was some randomization involved after 19 selecting patients who had already been treated 20 with antidepressant therapy. 21 Q. And had become suicidal while 22 being treated with antidepressants, correct? 23 A. I believe so. That makes 24 sense, logically, in a rechallenge sense, I can't Page 131 1 remember a specific wording in the protocol. 2 Q. Is it your understanding that a 3 study was being constructed or a clinical trial 4 was being constructed to test the validity of the 5 MSSIR? 6 A. Yes. 7 Q. To your knowledge, was that 8 study ever done? 9 A. I'm not sure. 10 Q. Why were you at this meeting on 11 May 13, 1991, meeting at the FDA discussing 12 rechallenge? 13 A. Most likely because I was a 14 manager or a leader of a group of statisticians 15 involved with the suicidality issue. And 16 secondly, I had been involved with Charles 17 Beasley, either before or after this meeting, and 18 actually talking about with Ivan Miller, I think 19 I was involved with a meeting that Ivan Miller 20 when he came on-site to Lilly had, so I had been 21 involved in that manner. 22 Q. So you met with Doctor Beasley 23 and Doctor Miller? 24 A. I've met with them, yes. Page 132 1 Q. On how many occasions? 2 A. I can't remember, at least one 3 where he came to Lilly. 4 Q. And what was the subject of 5 that meeting? 6 A. The MSSIR and reconfiguring it 7 and how to validate it. 8 Q. To your knowledge, was the MSSI 9 a scale that had been originally developed by 10 Doctor Miller and was being revised for the 11 rechallenge protocol at Lilly? 12 A. To the best of my knowledge. 13 Whether he developed it or not I'm not exactly 14 sure, but I think he was associated with it 15 somehow. Either he had used it or had developed 16 it, I'm not sure, but he was associated with it. 17 Q. Does the Beck scale ring a bell 18 with you? 19 A. I've heard of it but I do not 20 know what it assesses or how it assesses 21 patients. 22 Q. What was the status of the 23 validation study the last time you were aware of 24 what the status was? Page 133 1 A. The status was is that a 2 statistician in our group who was working with 3 Charles Beasley on that study, they had been 4 coming up with a protocol and I'm not sure 5 exactly where it went from there but they were 6 working on that protocol together. 7 Q. Who was the statistician? 8 A. Curtis Wiltse. 9 Q. Would you spell? 10 A. W-I-L-T-S-E. 11 Q. What was Mr. Wiltse doing with 12 regards to the validation protocol? 13 A. He, for example, did extensive 14 literature review just looking at different 15 scales, and I believe in this particular protocol 16 not only were they administering the MSSI or the 17 R revised but other scales as well to see how 18 well they corroborated with one another. So he 19 was just involved with doing background research 20 with the scales and helping Doctor Beasley design 21 the study. 22 Q. Was he setting up a statistical 23 analysis to be used with the data that was 24 collected through the validation protocol? Page 134 1 A. I believe he was. 2 Q. Are you familiar with what 3 analysis that was? 4 A. No, I'm not. 5 Q. What was your function with 6 regards to the development of the protocol for 7 the validation study? 8 As I mentioned, I worked with 9 Charles initially and then Curtis carried it on. 10 I just did some initial work and I'm not sure -- 11 well, I think most definitely Curtis started 12 working with Charles after this particular 13 meeting but at least up until this point in time, 14 I had been involved and then Curtis took it from 15 there. 16 Q. And this meeting occurred, the 17 meeting we are talking about in Exhibit 1, while 18 you were a research scientist? 19 A. It's -- I have to go back and 20 remember when I was promoted to manager. I 21 became a manager in '89, I believe, so I was a 22 manager at this point in time, a statistician but 23 in an administrative role. 24 Q. So would you have been superior Page 135 1 over Mr. Wiltse? 2 A. Yes. At that point in time he 3 did not report to me so I can't call myself his 4 superior, but organizationally I was a manager 5 and he was a senior statistician, so with respect 6 to salary, I was. 7 Q. Then why is it that he didn't 8 report to you? 9 A. We have a large staff and not 10 all the statisticians while I was a manager 11 reported to me, just a certain group of 12 statisticians, and Curtis was not one of those 13 that reported directly, he reported to a group 14 leader, Walt Dolphin and -- who reported to 15 Doctor Sampson, we all reported up to Doctor 16 Sampson. 17 Q. When was the last time you 18 heard about the status of the rechallenge 19 protocol? 20 A. I can't recall. 21 Q. How about the validation 22 protocol, when was the last time you heard about 23 the status of that? 24 A. I can't recall. Page 136 1 MR. SMITH: Was it done, either one? 2 THE WITNESS: Was the study done? 3 MR. SMITH: Yes. 4 Q. Were either studies done? 5 A. I'm not sure. 6 Q. Who would know that? 7 A. I'm not sure exactly who would 8 know that, I mean Charles Beasley was a physician 9 who while I was working with him and Curtis was 10 actually designing the study. 11 Q. Why is it you would have been 12 involved at the level where you spoke with the 13 FDA but you wouldn't be aware of what happened 14 with regards to whether or not the trials were 15 run? 16 A. I was involved initially and 17 then Curt Wiltse took over. 18 Q. Why? 19 A. As a manager, I had many 20 responsibilities and I'm involved with resourcing 21 studies properly and that was felt to be the best 22 way to resource this to have an experienced 23 statistician like Curt Wiltse working on that 24 particular study. Page 137 1 Q. Did you make the 2 decision to put Mr. Wiltse on the study? 3 A. I was involved with that 4 decision. 5 Q. To what extent? 6 A. I'm not sure exactly how Curt 7 was put on there but I was involved with just 8 saying we need someone who's experienced to work 9 on this, who's a good person to put on the study 10 and Curt was a definite option and we selected 11 him. So I'm not sure exactly whether I said I 12 want Curt or whether someone offered Curt up but 13 I was involved in that decision. 14 Q. Did you say earlier that at the 15 time that Mr. Wiltse was new to the company that 16 he was assigned to this project? 17 A. No. 18 Q. How long had he been with the 19 company? 20 A. I'm not sure. 21 Q. Had he been there longer than 22 say Vin Rampey? 23 A. I can't recall. 24 Q. Was Bruce Dornseif with the Page 138 1 company at this time? 2 A. I don't know when he resigned. 3 Q. How about Janet Bosomworth, was 4 she with the company at this time? 5 A. Yes. 6 Q. How about Mary Saylor, was she 7 with the company at this time? 8 A. I'm not sure as of May 15, 9 1991, whether she was in our department or not. 10 Q. Were any other senior 11 statisticians considered for to work on these 12 studies? 13 A. I'm not sure. Curt was the -- 14 I don't remember any disagreement on Curt working 15 at all, he was the solid choice because of his 16 experience and depth of knowledge. 17 Q. Was it your impression from 18 this meeting that the FDA wanted Eli Lilly to run 19 a rechallenge protocol on suicidality? 20 A. I'm not sure if they wanted 21 Lilly to do it, it's my impression that they 22 thought it was a good idea but I'm not sure who 23 initiated that concept. 24 (PLAINTIFFS' EXHIBIT NO. 2 WAS Page 139 1 MARKED FOR IDENTIFICATION AND 2 RECEIVED IN EVIDENCE.) 3 Q. Have you had a chance to look 4 at Exhibit 2? 5 A. Yes, I have. 6 Q. Is this the protocol for the 7 validation study of the MSSIR? 8 A. Yes. 9 Q. Does this refresh your 10 recollection as to whether or not this study was 11 actually performed? 12 A. No, it does not tell me whether 13 or not the study was actually performed or not. 14 Q. I believe earlier when I asked 15 you if you knew whether or not this study was 16 performed, you said you weren't sure. Do you 17 have any idea whatsoever whether or not it was 18 performed? 19 A. I'm not sure. 20 Q. You're not sure? 21 A. I know -- I see it and I know 22 this protocol was drafted, I'm not sure how far 23 it got with respect to actually enrolling 24 patients. Page 140 1 Q. Do you know if investigators 2 were hired to run the study? 3 A. I can't speak to that, I'm not 4 sure. 5 Q. Do you know if Doctor Miller 6 intended to be involved in running the study to 7 validate the scale? 8 A. Yes. 9 Q. Do you know if people working 10 with Doctor Miller were also going to run studies 11 or was it something that other people were going 12 to be hired to work on the studies outside of 13 Doctor Miller's group? 14 A. I'm not sure, I do note that it 15 says here multicenter so I might assume logically 16 that there were more to be included other than 17 Doctor Miller. 18 Q. Do you know how many people 19 were going to be enrolled in the study? 20 A. Under sample size criteria it 21 says maximum of two hundred and fifty patients. 22 Q. Did it have a minimum? 23 A. I don't know, in my brief 24 reading here I did not see a minimum but it gives Page 141 1 a maximum. 2 Q. Besides the validation study, 3 and the rechallenge protocol, were there any 4 other validation studies that you know of that 5 were going to be run? 6 MR. MYERS: Validation of what? 7 MS. ZETTLER: Validation of scales. 8 A. I have to go back a bit. I 9 can't recall a rechallenge protocol, per se, like 10 this. So you mentioned rechallenge protocol, I 11 don't know what to think about that. 12 Q. Besides the validation study 13 for the the MSSIR, are you aware of any other 14 validation studies that were going to be run? 15 A. No. 16 Q. Do you know if they were 17 intending on running a validation study on the 18 Hamilton depression rating scale? 19 A. No. 20 Q. Is Curt still with the company? 21 A. Yes. 22 Q. Do you know what he's working 23 on now? 24 A. In general, yes. Page 142 1 Q. Okay. Generally, what is he 2 working on? 3 A. He's working on products in the 4 cardiovascular area. 5 Q. He's no longer working with 6 Fluoxetine to your knowledge? 7 A. To the best of my knowledge, he 8 is not currently working on any Fluoxetine 9 projects. 10 Q. If this study, the validation 11 study had been run and data had been collected as 12 a result of this study, would you have been 13 involved in any way with the analysis of that 14 data? 15 A. Not with the analysis, that 16 again was under Curt's purview, but I may have 17 been involved with the review process to again 18 review reports. 19 Q. Have you ever been asked to 20 review a report resulting from a validation 21 study? 22 A. No. 23 Q. Have you ever been asked to 24 review a report as resulting from a rechallenge Page 143 1 study? 2 A. No. 3 Q. Other than the, let's call it 4 pre-Teicher studies that were run in 5 retrospective analysis of the clinical trials, 6 have you been asked to review reports for any 7 other studies? 8 MR. MYERS: What do you mean 9 pre-Teicher? 10 Q. Are you aware of Doctor Martin 11 Teicher? 12 A. Yes. 13 Q. Are you aware that Doctor 14 Teicher wrote an article that he published in 15 1990 that suggested that there was an increase in 16 incidence of suicidality with patients who used 17 Fluoxetine? 18 A. I'm familiar with an article he 19 wrote. 20 Q. Are you aware that that in 21 effect started the reanalysis of suicidal 22 ideation data on suicidality data by the FDA and 23 by Eli Lilly? 24 A. Yes. Page 144 1 Q. After that time, January of 2 1990, what types of studies were you asked to 3 review? We've already talked about the 4 information for the Beasley article in the 5 British Medical Journal, right? 6 A. Yes. 7 Q. And we've already talked about 8 the information that was being submitted to the 9 FDA Advisory Committee, right? 10 A. Yes. 11 Q. Anything else? 12 A. Other publications since then -- 13 Q. Okay. 14 A. -- dealing with Fluoxetine and 15 suicidality I've reviewed as well. 16 Q. Were all of those publications 17 based on retrospective analyses of data? 18 A. In other words, data that's 19 already been collected, completed studies and 20 just -- 21 Q. Similar to what was done with 22 the Doctor Beasley BMJ article? 23 A. Yes. 24 Q. Were any of those studies -- Page 145 1 have you ever worked on a prospective study set 2 up just to study suicidality, not necessarily a 3 rechallenge but a study that was set up to study 4 a suicidality rate within the population? 5 A. As I mentioned, the only work I 6 had initially was with Doctor Beasley in looking 7 at the validation. Beyond that, I have not been 8 involved in designing any such study. 9 Q. Have you ever worked on 10 suicidality articles as it relates to other 11 indicated uses of Fluoxetine? 12 A. I have, yes, I have worked on 13 one article that relates to suicidality and 14 adverse events, I forget the name of the title, 15 but that's one article where I'm an actual author 16 on that one. 17 Q. What's the indications used on 18 that article? 19 A. What kinds of patients are 20 included in that article? 21 Q. Right. 22 A. I'm not sure whether it's just 23 depressed patients or patients across many 24 indications, I can't recall right now. Page 146 1 Q. That's not the Beasley BMJ? 2 A. No, I was reviewer of that 3 publication. 4 Q. Do you remember what the 5 publication was of the article, the suicidality 6 article that you were a co-author on? 7 A. I'm sorry. 8 Q. Do you remember what 9 publication the article that you just told me 10 about -- 11 A. What journal it came out in? 12 Q. Right. 13 A. No, I do not. 14 Q. Do you remember when it was 15 published? 16 A. To tell the truth, I don't know 17 if it has been published yet. I've worked on it 18 and written a draft and worked with the authors, 19 I'm a co-author and I'm not sure whether that's 20 been published yet or not. 21 Q. Have you written a lot of 22 articles? 23 A. I've written a lot of articles 24 but not Fluoxetine articles. I've written a Page 147 1 number of articles dealing with statistical 2 methods. 3 Q. Have you written, besides that 4 one article, have you written any other 5 Fluoxetine articles? 6 A. I have been involved with an 7 article with bulimia studies, I believe I'm on at 8 least one if not more with respect to obesity. 9 Q. Okay. Any others with respect 10 to depression? 11 A. I am a co-author on an article 12 that spans all the indications, it's an article 13 titled Paradoxical Worsening and it looks into 14 depression plus obesity, bulimia, et cetera. 15 Q. And that's worsening of what, 16 depression? 17 A. Worsening of depression, 18 defined paradoxical worsening in that paper as 19 some unusual worsening, whether bulimia 20 phenomena, obesity, gaining a lot of weight back 21 or depression. 22 Q. And was that article published? 23 A. I believe so, yes, recently, I 24 believe 1993. Page 148 1 Q. Who else was author of that 2 article? 3 A. The paradoxical worsenings 4 article? 5 Q. Right. 6 A. Gary Tollefson, Vin Rampey, 7 myself, at least those three. 8 Q. Doctor Tollefson was the 9 research physician on the projects? 10 A. Yes, the lead author. 11 Q. I'm sorry, I may have just 12 asked you this and if I did, I apologize. Do you 13 remember what journal that was published in? 14 A. No, I don't. 15 Q. But your recollection is it was 16 published fairly recently within -- since the 17 beginning of 1993? 18 A. Yes, to the best of my 19 knowledge. 20 Q. Is this a retrospective 21 analysis of information? 22 A. Yes, this is again analysis of 23 the data base that we have. 24 Q. This wasn't based on any Page 149 1 retrospective study that would -- I'm sorry, any 2 prospective study that was set up specifically to -- 3 A. No, this is again a part -- 4 Q. Let me finish my question. 5 A. I'm sorry. 6 Q. It wasn't a result of a 7 prospective study measuring worsening across the 8 board in different indications? 9 A. No. 10 Q. This other article that we were 11 just talking about a few minutes ago that you 12 weren't sure whether or not it was published, 13 that was an analysis of suicidality? 14 A. Yes, and its relationship to 15 adverse events. 16 Q. Other adverse events? 17 A. Yes. 18 Q. Such as, what types of adverse 19 events, rashes? 20 A. One term that pops in mind 21 again, akathisia pops up in my mind as one 22 adverse event, that's all I can remember. 23 Q. How about tardive dyskinesia? 24 A. I'm not sure. Page 150 1 Q. How about dystonia? 2 A. I'm not sure. 3 Q. Who's the co-author on that 4 article? 5 A. Doctor Beasley, I believe, is 6 the research physician on that and co-authors 7 include myself and Vin Rampey again. 8 Q. When did you write a draft of 9 this article? 10 A. I don't know exactly when we 11 wrote the first draft, it was maybe a year and a 12 half ago. 13 Q. Do you know if the article has 14 been submitted for publication? 15 A. I believe it has been 16 submitted. 17 Q. Was it rejected? 18 A. No, I believe it came back for 19 review and, you know, the reviewer's comments 20 were incorporated and that's as far as I know 21 with the status of it. 22 Q. Do you know if there were 23 statistical problems with it as far as the 24 publication is concerned? Page 151 1 A. To the best of my knowledge, 2 no. 3 Q. What publication was it 4 submitted to? 5 A. I don't recall. 6 Q. Do you remember how long ago it 7 was submitted? 8 A. Resubmitted or submitted? 9 Q. Submitted initially. 10 A. No, I don't know. 11 Q. Do you know when it was 12 resubmitted? 13 A. No, I don't. 14 Q. Was it within the last six 15 months? 16 A. I can't say for sure. 17 Q. Do you remember the title of 18 the article? 19 A. No, I don't remember the title. 20 Q. Can you give me an idea? 21 A. Well, the words I do remember 22 are suicidality and adverse events, something to 23 that extent. 24 Q. Again was this a retrospective Page 152 1 study of clinical trial data? 2 A. It was analysis of data that we 3 already have in-house. 4 Q. Okay. It wasn't based on any 5 prospective study that was set up to study 6 akathisia or other adverse events in it as they 7 related to suicidality? 8 A. No, it was not. 9 Q. Is Doctor Beasley still working 10 on Fluoxetine issues as far as you know? 11 A. I'm not sure. 12 Q. When is the last time you 13 talked with Doctor Beasley? 14 A. I'm not sure -- maybe, I 15 believe -- I'm not sure. 16 Q. Was it more than a year ago? 17 A. No. 18 Q. Was it more than six months 19 ago? 20 A. No. I think within the past 21 month I said hello to him and that was it. 22 Q. How about actual substantive 23 conversations about Fluoxetine? 24 A. No, I have had no substantive Page 153 1 conversations with him about Fluoxetine within 2 the past year. 3 Q. Even with regards to the 4 akathisia article? 5 A. To the best of my knowledge, 6 no. 7 Q. What was your role with regards 8 to that article, were you the statistician 9 involved? 10 A. As a statistician, I worked 11 with Doctor Beasley to come up with a framework 12 by which to analyze the data and then the other 13 statistician, who I mentioned as a co-author, Vin 14 Rampey, actually implemented and refined it and 15 carried on the actual work with Charles Beasley, 16 so I was upstream a bit as far as the original 17 concept stage. 18 Q. When you'd work on a concept 19 say with Doctor Beasley or one of the other 20 clinical research physicians and you were going 21 to study say the incidence of akathisia within 22 the different trials, within the different 23 treatment groups, the comparitors, the 24 Fluoxetine, placebo, okay, who would make a Page 154 1 decision whether or not a specific incidence was 2 in fact say akathisia in this case? 3 A. I don't understand. 4 Q. Who would make a decision 5 whether or not any given patient within the group 6 was actually suffering from akathisia? 7 A. To the best of my knowledge, 8 the investigator who's actually caring for that 9 patient makes the determination and that 10 information is fed into the data base and then we 11 use that information that's in the existing data 12 base. 13 Q. Was that the same for the 14 suicidality analyses that you did for Doctor 15 Beasley's BMJ article? 16 MR. MYERS: Was what the same? 17 Q. Was that an analysis of what 18 was in the data base as a result of reports by 19 clinical investigators? 20 A. To my knowledge, that was at 21 least part of it. 22 Q. What was the other part? 23 A. Well, I do know that the 24 physicians within Lilly also helped comb through Page 155 1 every piece of data and came up with terms in 2 this data base they wanted to look for to make 3 sure they didn't miss anything. 4 Q. So let's talk about a Beasley 5 article for a couple of seconds. So you're 6 saying that to the best of your knowledge, 7 information that was reported by clinical 8 investigators were placed in the data base and 9 rereviewed or reviewed by clinical research 10 physicians at Lilly to search for event terms or 11 things of that nature, correct? 12 A. Yes, to the best of my 13 knowledge, that's true. I don't know the actual 14 process but I do know that both investigator 15 terms were used as well as internal review and 16 that's typical for many clinical trials and 17 that's why we have a team where the research 18 physicians are involved with scrutinizing the 19 data too. 20 Q. So that happens in any trial, 21 the clinical investigator will report his 22 findings made under his judgment to the company, 23 somebody at the company will review that 24 information to check for accuracy in whatever Page 156 1 test they want -- 2 A. For clarification, what exactly 3 does the person mean out in the field. 4 Q. And that information is placed 5 into a data base, correct? 6 A. Yes. 7 Q. And then in the case of Doctor 8 Beasley's article, the data was retrieved from 9 the data base or at least looked at on the data 10 base to make a determination if everything was 11 complete in there and whatever else? 12 A. There were specific 13 determinations made to define a patient as being 14 in this class of suicidality or not with respect 15 to time on drug or time elapsed after 16 discontinuing drug whereby they experienced the 17 event or not. I'm not sure exactly what those 18 cut points were but there were some 19 determinations made from a definition standpoint. 20 Q. There was also a determination 21 made regardless of time on drug as to whether or 22 not a specific act was actually suicidal gesture, 23 correct? 24 A. I'm not sure. I do know that Page 157 1 the physicians went through these algorithms to 2 determine whether the patient met the definition 3 or not. 4 Q. Were you involved in setting up 5 the algorithms for them to review? 6 A. No. 7 Q. Who was involved in doing that? 8 A. I'm not sure. The physicians 9 used them and I believe that physicians generated 10 them based on their clinical expertise. 11 Q. Is it your understanding that 12 they reviewed suspected suicidal ideation or 13 suicide attempt for the purpose of determining 14 whether or not it was actually related to 15 suicide, suicidal ideation? 16 A. I'm sorry. 17 Q. Let me ask it this way: Did 18 they look at every single case report form for 19 every single patient regardless of whether or not 20 they suffered an adverse event or did they narrow 21 it down to adverse events and then review them? 22 A. I'm not sure exactly what 23 process they went through, I know they looked at 24 a lot of case report forms. Page 158 1 Q. Do you know whether or not they 2 started out with a group of people who the 3 clinical investigators reported as suffering from 4 suicidal ideation and reviewed them narrowing 5 down the number? 6 A. I don't believe that was the 7 case, I believe they broadened their net wider 8 than what the investigators submitted but I don't 9 know the specifics on how broad that net was. 10 Q. Why do you believe they 11 broadened their net? 12 A. Why did they broaden their net? 13 Q. Why do you believe they 14 broadened their net? 15 A. I believe they broadened their 16 net to make sure if these phenomena were 17 occurring that they would find it. 18 Q. Who told you that, where did 19 you come to that understanding? 20 A. I came to that understanding 21 just in meetings with the physicians justifying 22 why they were using this certain algorithm. 23 Q. Did you ever request why they 24 were using the certain algorithms? Page 159 1 A. No. 2 Q. Why did they feel like they had 3 to justify using the algorithm? 4 MR. MYERS: Let me object to the form. 5 I don't know that he testified anybody had to 6 justify anything and it's speculation as to why 7 they did something. 8 A. They have an algorithm that 9 made clinical sense and they are telling people 10 here it is and here's why it makes clinical 11 sense. 12 Q. Who developed the algorithm? 13 A. I'm not sure. 14 Q. Would it have been a 15 statistician or clinical research physician? 16 A. I don't believe any 17 statistician was involved with developing the 18 algorithm. 19 Q. Would it have been a clinical 20 research physician as opposed to somebody else? 21 A. It was a matter of clinical 22 determination and hence I would believe it was 23 someone with that background, a research 24 physician or other physician, but I'm not sure Page 160 1 explicitly who came up with it. 2 Q. Have you ever done an analysis 3 of the raw data in the data base as a result of 4 the double blind controlled studies as opposed to 5 the data that was given to you by the clinical 6 research physicians? 7 A. What do you mean by raw data? 8 Q. Well, it's my understanding 9 that the data from the clinical trials was stored 10 on data bases in say the System 38 and then it 11 was transferred to the main frame for analysis 12 with SAS, is that correct? 13 A. System 38 was included, it may 14 not all be there but it was included in that but 15 ultimately it gets into SAS for analysis, that's 16 correct. 17 Q. So at some point all the data 18 was transferred to SAS for analysis, it ended up 19 in SAS, it ended up in the data base so it could 20 be analyzed through SAS? 21 A. Yes. 22 Q. And after the research 23 physicians looked at the algorithms, a 24 determination was made as to what cases would fit Page 161 1 into the criteria that they set forth for 2 suicidality, be it time on drug or whatever other 3 parameters they used, correct? 4 A. Yes. 5 Q. So logically either more cases 6 were thought to be suicidal, suicidality-related, 7 or less, depending on how wide they cast their 8 net, right? I mean it could have been that the 9 investigators may have, for example, cited ten 10 cases to look for suicidal ideation, okay, and 11 after they reviewed the data, the clinical 12 research physicians decided that there may be 13 twelve that were really related to suicide, 14 correct? 15 A. That's possible, I'm not sure 16 what the numbers were. 17 Q. Have you ever done an analysis 18 of the data that was in the data base to be 19 analyzed before the research physicians made 20 their analysis with the algorithm? 21 A. I personally have not done any 22 analysis, per se, of the data base. 23 Q. To your knowledge, has anybody 24 done such an analysis? Page 162 1 A. There may have been. 2 Q. Who would have done that, if 3 anybody? 4 A. I'm not sure. I do know that 5 there's been analysis of these things before, 6 just as far as the safety updates by Bruce 7 Dornseif who was the statistician involved, and 8 he may have done an analysis before the algorithm 9 came in place, but I'm not sure. 10 Q. They would report raw 11 percentages, in other words, and in a hundred 12 cases out of a thousand clinical investigators 13 reported that somebody suffered a certain kind of 14 rash while on Fluoxetine? 15 MR. MYERS: You mean in a safety 16 update? 17 MS. ZETTLER: Right. 18 A. That's one way that they report 19 the data. 20 Q. They wouldn't do an analysis of 21 the incidence of -- incidence of that rash across 22 treatment groups, would they, for safety updates? 23 A. Yes, they might compare the 24 actual treatment groups on the incident rate. Page 163 1 Q. And again that would be 2 something that -- whether or not a person had 3 actually suffered a rash would be a determination 4 made through an algorithm? 5 MR. MYERS: Are we talking about safety 6 update? 7 A. For the rash? 8 Q. Right. 9 A. Well, there's a definition, 10 whether you want to call that algorithm or not 11 there's a definition of rash, and so, yes. 12 Q. And that definition is made by -- 13 or that determination is made by the clinical 14 research physician? 15 A. The investigator, you know, the 16 one taking care of the patient, observes the rash 17 and then that is fed into the data base and so 18 it's the investigator actually who makes the 19 initial determination. 20 Q. Have you ever seen a listing of 21 incidence of suicidality and suicide or suicide 22 attempts as reported by clinical research 23 physicians? 24 A. I can't say that I have. Page 164 1 Q. Can you say that you haven't? 2 A. No, I can't recall if I have 3 seen such a thing. 4 Q. Are you aware of whether such a 5 list was ever compiled? 6 A. I'm not certain if such a list 7 was compiled or not. 8 Q. Did the clinical research 9 physicians make a determination as to whether or 10 not a particular act was related to suicidality 11 on their own or did they have consultants to help 12 them out? 13 MR. MYERS: For the purposes of this 14 reanalysis you were talking about? 15 MS. ZETTLER: For the purposes of 16 anything. 17 A. Could you rephrase the question 18 again? 19 Q. Sure. It's your understanding 20 that there were more than one analyses done on 21 the clinical trial information with regards to 22 Fluoxetine and suicidality, correct? 23 A. There have been many analyses. 24 Q. And to your knowledge, has Page 165 1 Lilly consulted with anyone outside of Lilly with 2 regards to analyzing the data regarding 3 suicidality to be used then in an analysis of the 4 incidence of suicidality in Fluoxetine? 5 A. Yes, Lilly has used 6 consultants. 7 Q. On how many occasions? 8 A. On how many occasions what? 9 Q. Have they consulted with people 10 outside? 11 A. I wouldn't know how many, I 12 know there's been multiple times. 13 Q. More than ten? 14 A. Oh, I wouldn't know because 15 that might involve a phone call, it might involve 16 bringing them into Lilly, I'm not sure how many 17 actual contacts have been made. 18 Q. I'm not talking about contacts, 19 I'm talking about actual people, numbers of 20 people, how many are you aware of? 21 A. Numbers of people, oh, I'm not 22 aware of a number greater than ten. 23 Q. Can you give me some of those 24 names? Page 166 1 A. Well, from a statistical 2 standpoint, we used one consultant just to help 3 us with the methods we used, Skip Wilson. 4 Q. Anybody else? 5 A. From the clinical standpoint, I 6 can't remember any of the names of the physicians 7 that some of the research physicians have used to 8 consult with. 9 Q. How about Jan Fawcett, does 10 that name sound familiar? 11 A. That name sounds familiar but 12 I'm not sure what he did with our research 13 physicians. 14 Q. You don't recall any of the 15 names of the consultants from a clinical 16 standpoint? 17 A. Jan Fawcett rings a bell, he 18 may or may not have -- well, that name rings a 19 bell. 20 Q. Anybody else, anybody that you 21 can remember? 22 A. No. 23 Q. Would you consider Doctor 24 Miller, Ivan Miller a consultant with regards to Page 167 1 suicidality? 2 A. For that particular -- for the 3 prospective validation study, I know he was a 4 consultant. Whether he consulted with the 5 research physicians on the retrospective 6 analysis, I can't say, I don't know. 7 Q. Was that one time Doctor 8 Tollefson, was he a consultant? 9 A. I believe so. He had some 10 relationship with Lilly before he was hired by 11 Lilly. Whether he was, quote, a consultant or 12 not, I'm not sure how they used him but he had 13 some relationship there. 14 Q. Any other consultants that you 15 can think of regardless of whether or not they 16 were related to clinical? 17 A. No, again my dealings were 18 mostly with Skip Wilson on the statistical side. 19 Q. Why did you use Skip Wilson as 20 opposed to just doing everything in-house? 21 A. We wanted someone outside of 22 Lilly who was respected in the statistical 23 community and he's also respected in the field of 24 meta analysis and providing different studies Page 168 1 together and we knew of some publications that he 2 had that made us believe that he knew what he was 3 talking about and so we called him up and he 4 agreed to look at our methods with us and help 5 steer us with respect to statistical methodology. 6 Q. With the Beasley article in 7 particular or any other problem? 8 A. Beasley article in particular 9 but that model was used across a number of 10 publications. 11 Q. It is my understanding that the 12 Beasley article is based solely on seventeen U.S. 13 double blind controlled studies, correct? 14 A. That's my impression. 15 Q. It's also my understanding that 16 the article was original submitted to the New 17 England Journal of Medicine but it was turned 18 down, is that correct? 19 A. I can't state whether I know it 20 was submitted or not to the New England Journal. 21 I heard the name New England Journal but I don't 22 know if it was ever submitted to them or not. 23 Q. Did they ever come back to you 24 and say Greg, we submitted this to any Page 169 1 publication and it's been turned down because of 2 problems with the statistical analysis? 3 MR. MYERS: The Beasley paper? 4 MS. ZETTLER: Yes. 5 A. No. 6 Q. Did you have problems with the 7 British Medical Journal with regards to the 8 statistical analysis of data on that article? 9 A. No. 10 MS. ZETTLER: Do you want to break now? 11 It's a good place to stop. 12 MR. MYERS: That will be fine. 13 * * * * * * * * * * 14 (THE DEPOSITION WAS ADJOURNED UNTIL 15 THE FOLLOWING MORNING, 9-17-93 AT 16 9:00 A.M. AT THAT TIME, THE 17 PROCEEDINGS CONTINUED, AS FOLLOWS:) 18 * * * * * * * * * * 19 20 Q. (BY MS. ZETTLER) Doctor, you 21 understand that you're still under oath? 22 A. Yes. 23 Q. We don't need to go through 24 that whole rigamarole again. I think when we Page 170 1 left off yesterday we were talking about the BMJ 2 article and whether or not the publication had 3 any problems with the statistical analyses and 4 things of that nature. Do you remember that? 5 A. Yes. 6 Q. And I think your answer was 7 that they did not have any problem with it? 8 A. No. 9 Q. No, they didn't have a problem 10 or no -- 11 A. No. 12 Q. No, they did not have a problem 13 with it. 14 A. We did not have a problem with 15 the statistical analysis. 16 Q. Okay, but how about the 17 magazine, the Journal, did they have a problem 18 with the statistical analysis? 19 A. No. What we submitted 20 originally was ultimately published. 21 Q. Was there a period of time when 22 they asked you to redo any of the statistical 23 analyses and then it went back to the original 24 analysis? In other words, did you submit it -- Page 171 1 when it was submitted, was the first submission 2 ultimately published or published as opposed to a 3 revision of the first submission you made to the 4 BMJ? 5 MR. MYERS: The statistics or -- 6 MS. ZETTLER: The entire article. 7 A. I'm not sure whether there was 8 a revision or not. I believe there may have 9 been, but I'm uncertain how many times there were 10 revisions, but that was for the the entire 11 article. The original statistical methodology 12 that we had developed and used for the original 13 document was ultimately published. 14 Q. Okay. How about the New 15 England Journal of Medicine, do you remember if 16 they had a problem with the statistical analysis? 17 A. I don't recall, I'm very fuzzy 18 about the New England Journal aspect. 19 Q. Did you have a problem with the 20 statistical analysis used in the BMJ article? 21 A. No. 22 Q. Did you have a problem with how 23 the data was prepared to be analyzed? 24 A. No. Page 172 1 (PLAINTIFFS' EXHIBIT NO. 3 WAS 2 MARKED FOR IDENTIFICATION AND 3 RECEIVED IN EVIDENCE.) 4 Q. Take your time and read through 5 the whole thing because I'm going to ask you a 6 bunch of questions about it. 7 A. Okay. 8 Q. Have you had a chance to review 9 it? 10 A. Yes. 11 Q. Are you sure? You can take 12 more time if you need it, that's okay. 13 A. I may need to as you ask 14 questions. 15 Q. Does this refresh your 16 recollection as to any problems or questions BMJ 17 had with regards to statistical analysis used in 18 the Beasley paper? 19 A. Yes. 20 Q. Can you tell me what kind of 21 concerns they had? 22 A. The reviewer was not clear in 23 the original paper as to how the methods were 24 applied. Page 173 1 Q. Why was he not clear? 2 MR. MYERS: Before he answers, let me 3 object to the form only to the extent you're 4 asking him what some reviewer at the British 5 Medical Journal thought. He can certainly 6 testify to his understanding, his understanding. 7 MS. ZETTLER: I think it's pretty well 8 documented in the exhibit, Larry, if he needs to 9 refresh his recollection. 10 Q. Well, let me ask you this: Do 11 you recognize this exhibit? 12 A. Yes. 13 Q. And you worked with this 14 exhibit at some point in helping clarify things 15 for BMJ, is that true? 16 A. I worked with our statisticians 17 who were doing the analysis. 18 Q. But you reviewed this exhibit 19 at some point in time? 20 A. Yes. 21 Q. Did you have any question as to 22 why the statistician at BMJ was unclear as to 23 what some of the analyses were? He said he was 24 unclear, right? Page 174 1 A. Yes, he states here he's 2 unclear. 3 Q. Can you understand from this 4 exhibit why he was unclear, does he state what 5 things were unclear to him? 6 A. Yes. 7 Q. Okay. 8 A. In particular, he gave this 9 Table A to him, it was something he proposed as 10 something that would make the data base very 11 clear for him. 12 Q. Were there any other problems 13 that he had with the article in general? 14 A. I can only speak to what this 15 draft says, I'm not sure all -- 16 Q. According to the draft, what 17 other problems or questions did he have with the 18 article? 19 A. Would you like me to go through 20 here and articulate them? 21 Q. Yes, please. 22 A. Okay. I mentioned this first 23 data picture, he suggests as a way to become 24 clear as to all the studies that were involved in Page 175 1 the analysis. 2 Q. And you're talking about? 3 A. Table A. 4 Q. Table A. That starts on page 5 Pz 2465 1429? 6 A. Yes. 7 Q. And it goes on to the next 8 page, Pz 1465 1430, right? 9 A. Yes. That was his first 10 suggestion. 11 Q. Let's stop there. Why did he 12 suggest this table? 13 A. I assume that to him it made 14 the data clearer. 15 Q. What data? 16 A. The data that was being 17 analyzed from the clinical trials. 18 Q. Okay. You really mean your 19 reporting of the data or you mean Lilly's 20 reporting of the data? He didn't have access to 21 the data in the data base, did he? 22 A. I don't believe so. 23 Q. So when you say made the data 24 clearer, you mean the data that you presented or Page 176 1 was presented in the original draft of the paper 2 or the original submission of the paper? 3 A. Yes. 4 Q. Go ahead, I'm sorry. 5 A. Likewise he suggests on Table B 6 and Table C some other analyses. 7 Q. What other analyses? 8 A. Where not only is the incidence 9 of suicidality displayed, but patient years, in 10 other words taking into account patient exposure. 11 Q. How long they were on the drug? 12 A. That's right. 13 Q. Be it placebo, tricyclic or 14 Fluoxetine or only Fluoxetine? 15 A. Fluoxetine, placebo and 16 tricyclic, as shown in Table B. 17 Q. And that's on page Pz 2465 18 1431, right? 19 A. Yes. 20 Q. What was it about Lilly's data, 21 as represented in the original submission, that 22 was unclear to him that he felt he had to redo 23 the table? 24 MR. MYERS: Before he answers, Nancy, Page 177 1 you're asking him to speculate as to what the 2 reviewer intended, meant or understood other than 3 what's in the letter, and the letter does speak 4 for itself. If he has any independent 5 understanding, certainly he can tell you, but 6 you're asking him to speculate what the reviewer 7 thought about the presentation of certain data. 8 MS. ZETTLER: No, I'm asking him to 9 explain to me from this exhibit his impression of 10 what the reviewer thought was wrong and what he 11 did to correct it. 12 MR. MYERS: I object to the form. If 13 you can answer the question, go ahead. 14 A. I cannot speak to the 15 reviewer's impression, I can only speak to my 16 impression of what he says here. 17 Q. That's fine. 18 A. And based on those tables, he 19 felt that it would be good to put patient years 20 as a column here to not only look at incidence 21 and take into account possible differential 22 treatment length for exposure, so that's another 23 column he suggested adding there. And I'm not 24 sure if we had previously done analyses using Page 178 1 patient years or not, I'm not sure about that, 2 but his suggestion was to put it right up front 3 there with the incidence rates. 4 Q. Why, if you can tell from this 5 exhibit, why did he feel that was necessary? 6 MR. MYERS: Same objection as to the 7 form, but if you can answer, go ahead. 8 A. He mentions under specific 9 comments, one, the author should choose between 10 binomial and life table rates. 11 Q. What does that mean, what is a 12 binomial? 13 A. A binomial is an incidence 14 rate, the straight incidence, how many patients 15 had an event divided by the total number of 16 patients. The life table rate is the patient 17 exposure analysis. So he's saying instead of 18 presenting or showing both of those, he's saying 19 the author should choose one or the other. 20 Q. Were both shown in the original 21 submission to the best of your recollection? 22 A. I'm not sure. I assume that 23 they were just because he's saying you have to 24 choose between the two, but I can't remember. Page 179 1 And then he mentions he prefers the latter. 2 Q. The binomial? 3 A. The life table rates, I prefer 4 the latter. 5 Q. I'm sorry, which page are you 6 referring to? 7 A. This is Pz 2465 1431, specific 8 comments, number one. 9 Q. Okay. He goes on to say 10 jumping from one to the other causes major 11 difficulties for the reader? 12 A. Right. 13 Q. Is it your recollection that a 14 combination of the binomial and life table rates 15 were used in the original submission? 16 A. Again, I don't remember 17 specifically, but I believe reading these 18 comments here that they may have been both 19 presented there, but I'm not certain. 20 Q. Okay. To your recollection, 21 the published form of the article, did you switch 22 to the life table rates? 23 A. I believe -- I know the 24 binomial rates, the incidence rates, were shown Page 180 1 there as well, and the life table rates may not 2 be shown in the BMJ article, I'm not sure. 3 Q. What other problems did he have 4 with the article? 5 A. Under specific comments, number 6 two, he thinks the title should be modified. 7 Q. For what reason? 8 A. I'm not -- I can't address that 9 because I'm not certain what the original title 10 was -- well, unless this is the title given on 11 the cover letter here. 12 Q. Where it says on the first page 13 of the exhibit, Fluoxetine and suicidality 14 absence of association in controlled depression 15 trials? 16 A. Yes, if that is the original 17 title. 18 Q. It may have been chained to the 19 meta analysis title that was ultimately used? 20 A. Possibly. I'm assuming, based 21 on this cover letter, that that was the title 22 that he saw when he reviewed the article. So he, 23 in specific comment point two, is questioning, I 24 believe, the use of the term absence of Page 181 1 association. 2 Q. Okay. And beginning of 3 specific comment two, he says clearly suicidal 4 acts are rare, in paren, zero point thirty-three 5 percent, close paren, in a multitude of patients, 6 and the total number of events observed, paren, 7 ten, close paren, is low, it is therefore not 8 surprising that a lack of statistical 9 significance is found between treatment groups; 10 correct? 11 A. Yes. 12 Q. Then he goes on to say, 13 however, a simplistic analysis -- and he refers 14 above to the correct approach; correct? 15 A. Yes. 16 Q. Gives oh hours of one point two 17 nine and one point seven oh for comparisons one 18 and two respectively, suggestive of an excessive 19 risk with Fluoxetine. What is he talking about 20 there? 21 MR. MYERS: I object to the form 22 because you're calling on him to speculate as to 23 what the writer means beyond what's in the 24 letter. Page 182 1 Q. Can you understand what he 2 means in that paragraph, Doctor? 3 A. I believe I understand the gist 4 of it. 5 Q. Okay. Tell me what the gist of 6 it is. 7 A. I have to go back. 8 Q. That's fine, take your time. 9 A. I'm not clear just looking at 10 this how he calculates odds ratios of one point 11 two nine and one point seven oh. He says 12 comparison Roman numeral one and two, but there's 13 ones and twos -- there's multiple Roman numeral 14 ones and twos there, so I'm not exactly sure how 15 he calculates those odds ratios, that's what O-R 16 stands for. 17 Q. Okay. He recalculated the odds 18 ratios and came up with a different number than 19 you guys did, though, didn't he? 20 A. We did not use odds ratios in 21 our paper here. 22 Q. Why not? 23 A. Our methods of analysis 24 compares the incidence rates in each group, and Page 183 1 the odds ratio is a different statistic than the 2 actual difference between the rates. 3 MS. ZETTLER: Can you read that back? 4 (THE COURT REPORTER READ BACK THE 5 REQUESTED TESTIMONY.) 6 Q. So he used a different 7 statistical analysis on the data and came up with 8 different numbers; correct? 9 A. In his specific comment one 10 point two, he says a simplistic analysis gives 11 odds ratios, and yes, we did not use the odds 12 ratios in our analysis. 13 Q. So the answer to my question is 14 yes, he did a different statistical analysis on 15 the numbers and came up with different numbers 16 than you guys did? 17 MR. MYERS: I object to the form 18 because they're two different analyses, so how 19 can the numbers be different, different than 20 what? 21 Q. He suggests in the paragraph 22 that the numbers were higher suggesting an excess 23 risk with Fluoxetine, quote, unquote. My 24 question is: Did he do a separate statistical Page 184 1 analysis on the data? 2 A. Yes. 3 Q. And he came up with different 4 numbers than you guys did, right? 5 A. I cannot say they're different 6 numbers because we didn't have the numbers. They 7 weren't different, he calculated the numbers, the 8 odds ratios. 9 Q. In his opinion, his numbers 10 suggest there's an excess risk with Fluoxetine, 11 that's what he says in that paragraph, right? 12 A. Yes. 13 Q. In fact, the excess risk caused 14 him to ask you to change the title of the paper, 15 right? 16 A. Yes. 17 Q. And the title of the paper, as 18 far as we know, resulted in this exhibit as 19 absence of association in controlled depression 20 trials; correct, that portion of the title? 21 A. Let me continue reading one 22 sentence here and then -- 23 Q. Sure. 24 A. Okay. Would you repeat that? Page 185 1 Q. Sure. He felt, through his 2 statistical analysis, that there was a suggestion 3 of an excess risk with Fluoxetine; correct? 4 A. Yes. 5 Q. And he was looking at the 6 suicidal acts rates; correct? 7 A. Yes. 8 Q. Okay. And because of his 9 findings through his statistical analysis, 10 suggesting an excess risk of suicide rates with 11 Fluoxetine, he suggested that you change the 12 title from Fluoxetine and suicidality absence an 13 association in controlled depression trials to 14 something that reflected what he found in his 15 statistical analysis; correct? 16 A. He didn't state specifically it 17 should reflect his particular conclusions, but he 18 did suggest the title should be modified. 19 Q. Changed from absence of? 20 A. Yes, that's my impression, 21 that's what he was focusing on. 22 Q. The next, point three, he says 23 the analysis of changing Ham-D does not seem 24 entirely appropriate; correct? Page 186 1 A. Yes. 2 Q. And what does he feel is not 3 appropriate about the analysis of Ham-D? 4 MR. MYERS: I object to the form on the 5 same basis. If he knows independently, he 6 certainly can tell you. 7 Q. If you can tell from the 8 exhibit why he feels that the analysis of the 9 change in the Ham-D is not entirely appropriate, 10 please tell me. 11 A. I'm not clear from this exhibit 12 because he says see figure, and I don't know 13 which figure he's referring to here. 14 Q. Can you take a couple of 15 seconds and look through the exhibit and see if 16 you can figure out what he means by the figure? 17 A. There's nothing here, as far as 18 I can tell, that's labeled figure. 19 Q. Let me ask you this: He says 20 in the next sentence, for example it is not 21 correct to make a comparison between treatment 22 groups for base line values pooled over all 23 trials. Does this indicate that all the trials 24 that were analyzed were pooled, the data from all Page 187 1 the trials were pooled before the Ham-D scores 2 were analyzed? 3 A. Yes. 4 Q. When you say pooled, what do 5 you mean, do you mean Fluoxetine versus all 6 comparitors or Fluoxetine versus each comparitor 7 individually? 8 MR. MYERS: Are you asking him what was 9 done as opposed to what the person means? 10 MS. ZETTLER: I'm asking him what was 11 done in the original analysis of the paper, 12 right. 13 A. There were seventeen trials in 14 the paper. 15 Q. Okay. 16 A. And as the Table A suggests, 17 there are different levels of pooling. 18 Q. What are those different 19 levels? 20 A. One level is to take the 21 Fluoxetine and placebo trials only and look at 22 those as a group because those trials only had 23 Fluoxetine treated patients and placebo treated 24 patients. Page 188 1 Q. Okay. 2 A. There's another set of trials 3 which include Fluoxetine and tricyclics, so 4 there's a pooling where you just combine and look 5 at those data. 6 Q. Okay. 7 A. And then there's another set of 8 trials that includes both Fluoxetine treated 9 patients, placebo treated patients and tricyclic 10 treated patients. 11 Q. When you say pooled, you mean 12 all the numbers are put together before they're 13 analyzed? 14 A. I'm saying those data alone are 15 analyzed together using appropriate statistical 16 methods. So right there I've mentioned three 17 different classes of studies that can be grouped 18 together and looked at. 19 Q. Okay. So when you say that 20 they're grouped together, do you mean all of the 21 Ham-D scores or all of the Ham-D data is grouped 22 together for all those things, Fluoxetine, 23 placebo and tricyclics, say, in the third data 24 pool that you were talking about, are all of Page 189 1 those put together and analyzed or are they each 2 analyzed individually within treatment group? In 3 other words, do you take the three groups, 4 Fluoxetine, tricyclics and placebos, in that one 5 case, and analyze everything together or do you 6 analyze them all separately compared to each 7 other? 8 A. The statistical methods -- I 9 believe there's two trials of that particular 10 nature, two three-arm trials. 11 Q. Okay. 12 A. The statistical methods used 13 compare the data within each one of those trials 14 and then it sums together the comparison within 15 each study. So it takes into account the two 16 separate studies that are involved in the 17 combined data analysis. 18 Q. So the data from both studies 19 are taken into account and analyzed together? 20 A. Yes. 21 Q. What does it mean to stratify, 22 what does stratify mean? 23 A. Stratify is to look at each one 24 of those studies, as I just suggested, the data Page 190 1 is compared within each one of those studies, and 2 then your overall statistical analysis takes each 3 of the per study comparisons and combines them 4 together. So a stratification refers to each 5 individual study taken on its own merits, and 6 then aggregate it together. 7 Q. So in other words, you have 8 another step that's involved, you analyze each 9 individual study and then you pool everything 10 together and analyze it? 11 A. Yes, it's all part of one 12 process, but that's generally the flow. 13 Q. Did you stratify the different 14 treatment groups for the Beasley article or did 15 you pool everything together? 16 A. No, we do not stratify based on 17 treatment group, we stratify based on the 18 particular study. 19 Q. Did you use a stratification in 20 analyzing the data for the Beasley article? 21 A. Yes, we used stratification per 22 the different studies involved. 23 Q. When did you do that, was that 24 done right off the bat? Page 191 1 A. I'm not sure initially when we 2 started using the method that was ultimately used 3 in the British Medical Journal paper. We had 4 used stratification in reporting results to the 5 Food and Drug Administration before the British 6 Medical Journal publication. 7 Q. Let me ask you this kind of as 8 an aside: Why wasn't the international study 9 data used in the Beasley paper, why did you limit 10 it just to the United States trials? 11 A. Statistically I cannot speak to 12 that. The physicians and others who know about 13 running international trials have reasons why 14 those data were included here. But from a 15 statistical perspective, we'll take data that's 16 felt clinically that possibly, though, should not 17 be included here. But as a statistician, I 18 cannot address all the reasons, I'm not sure of 19 those right now. 20 Q. Do you have any idea of why, 21 did anybody ever tell you why they were excluding 22 the international trials from the Beasley paper? 23 A. The only thing that sticks out 24 in my mind initially is just that those trials Page 192 1 were possibly different from the trials conducted 2 in the United States. 3 Q. Possibly different? 4 A. Possibly. As far as patients -- 5 you know, what kinds of patients were being 6 enrolled. And again, when you're doing combined 7 meta analysis, as I suggested yesterday, you want 8 to make sure you're comparing apples and apples, 9 and, so, there's possibly some concerns that all 10 that together might be not pure apples and apples 11 comparisons. But again, those are medical 12 issues, and I wouldn't have the full medical 13 justification. 14 Q. The people that were -- the 15 seventeen trials that were relied upon ultimately 16 in the Beasley article, people in those trials 17 were different, it wasn't technically apples to 18 apples, was it? 19 A. It was -- again, that's a 20 clinical question, so as a statistician, you 21 know, we use the clinician's judgment in talking 22 about patient demographics and how they're 23 treated. But basically it was felt that these 24 patients represented in this cohort of seventeen Page 193 1 studies were basically comparable. And when you 2 do statistics looking, for example, at base line 3 demographics and just looking at what kinds of 4 patients were in those trials, it appears that we 5 had a situation where we do have an apples and 6 apples comparison. 7 Q. Seventeen studies that were 8 included in the Beasley paper had different entry 9 criteria and different scales that were used, 10 things of that nature, didn't they? 11 A. I can't speak to how different 12 they were, I'm sure there were differences just 13 because, for example, there's obviously different 14 treatment groups being used in different studies. 15 So I'm sure there were, I don't know what those 16 differences actually were. 17 Q. You were in favor of 18 stratifying the data for the Beasley article, 19 weren't you? 20 A. Yes, we stratify -- all of our 21 trials are stratified based on center or 22 different investigators. So this is not unusual. 23 We can take a prospective clinical trial and 24 stratify there as well, based on centers, so this Page 194 1 is a standard statistical routine that is used. 2 Q. The answer to my question is 3 yes, right? 4 A. Yes. 5 Q. You were in favor of 6 stratifying the data to analyze the Beasley 7 article; correct? 8 A. Yes. 9 Q. And Doctor Beasley was more in 10 favor of pooling the data, wasn't he? 11 A. I'm not sure. 12 Q. Lumping the studies together, 13 he was more in favor of doing that as opposed to 14 stratification, wasn't he? 15 A. I can't recall exactly what he 16 was. He obviously is an author on this paper 17 where stratification was used. 18 (PLAINTIFFS' EXHIBIT NO. 4 WAS 19 MARKED FOR IDENTIFICATION AND 20 RECEIVED IN EVIDENCE.) 21 Q. Does this refresh your 22 recollection as to whether or not there was a 23 disagreement between you and Doctor Beasley as to 24 how the data should be analyzed? Page 195 1 A. It reflects differences in the 2 way data can be analyzed. 3 Q. It reflects a disagreement 4 between you and Doctor Beasley on how the data 5 should be analyzed? 6 A. No, not in particular. 7 Q. What do you mean where you say 8 at the bottom that it's obvious he wants the last 9 say in this, as I and others have stated it's not 10 primarily a stat question or a question of 11 numbers when one starts to think about pooling, 12 first you have to decide if the studies are 13 similar, we now believe, based on the new info 14 that the studies are not as similar as once 15 thought. What does that mean? 16 A. Again, it's a clinical question 17 with respect to the differences between these 18 kind of studies. So as a statistician, I can 19 only speak to analyzing the numbers from the 20 studies that are completed and given to us. So 21 the physician needs to have the final say with 22 respect to the differences between studies, why 23 we're including these studies and why we're not 24 including these studies. So we work with the Page 196 1 physician. 2 Q. Doctor Beasley had the final 3 say as to how you would analyze the data, pooling 4 it as opposed to stratification? 5 A. No. 6 Q. Who had the final say? 7 A. It was decided well up front 8 that these seventeen studies would be included in 9 this data analysis. 10 Q. Okay. 11 A. And if my memory serves me 12 correctly, I believe in the British Medical 13 Journal paper we presented more than one analysis 14 of the data, we presented a stratified analysis 15 and we presented an analysis without 16 stratification. 17 Q. So now you remember what 18 analyses were used in the paper? 19 A. Yes. I mean I know both were 20 presented there, I believe. 21 Q. Does this refresh your 22 recollection as to whether or not you used a 23 binomial or life table? 24 A. Whether we used stratification Page 197 1 or not, both used binomial rates, that's a 2 separate question. 3 Q. Okay. So you used binomial as 4 opposed to life table rates? 5 A. No, I'm not sure. I know we 6 used binomial, and we may or may not have also 7 included life table rates in the British Medical 8 Journal. 9 Q. What did you mean in the first 10 paragraph when you said one new factor I just 11 heard of today that influences the way I think 12 about lumping all studies together without 13 stratifying by study. What did you mean by that? 14 A. That's referring to the sixth 15 line down, these include different diagnostic 16 criteria, scales, exclusion criteria, et cetera. 17 Q. So the different studies 18 weren't as similar as you originally had thought. 19 A. That's correct. 20 Q. This is dated August 1, 1991, 21 isn't it? 22 A. Yes. 23 Q. This is after the revision 24 reflected or the suggested revision reflected in Page 198 1 Exhibit 3, which is dated July 22, 1991; correct? 2 A. Yes. 3 Q. Was this data or the 4 differences between the studies reanalyzed in 5 response to the statistician from BMJ's request 6 or suggestions? 7 A. The title to this paragraph 8 here, August 1, 1991, is some feedback on the BMJ 9 paper. So I assume it's related to this 10 discourse here. 11 Q. Okay. In the second paragraph 12 where you start talking about -- by the way, you 13 don't like capital letters, do you, very much. 14 The second paragraph that starts these include 15 different diagnostic criteria, scales, exclusion 16 criteria, et cetera, so I am now more convinced 17 that study stratification is the only way to go. 18 A. Yes. 19 Q. So you disagreed with pooling 20 the data at that point; correct? 21 A. No, not necessarily. 22 Q. Okay. Why not? 23 A. Stratification is the only way 24 to go for the primary statistical analysis. As Page 199 1 the BMJ paper put in, there is a pooling analysis 2 as well. But as the primary analysis, 3 stratification is the only way to go. 4 MS. ZETTLER: Larry, can I get a copy 5 of this, I didn't have a chance to make a copy 6 this morning. 7 (A SHORT RECESS WAS TAKEN.) 8 (PLAINTIFFS' EXHIBIT NO. 5 WAS 9 MARKED FOR IDENTIFICATION AND 10 RECEIVED IN EVIDENCE.) 11 Q. Doctor, have you had a chance 12 to review Exhibit Number 5? 13 A. Yes. 14 Q. This is a copy of the article 15 that was ultimately published in the British 16 Medical Journal authored by Doctor Beasley; 17 correct? 18 MR. MYERS: That's part of it. 19 MS. ZETTLER: Right. 20 MR. MYERS: There's some other 21 attachments. 22 MS. ZETTLER: Right. 23 Q. The first approximately eight 24 pages, I believe, is the Beasley article that Page 200 1 we've been talking about for the past half hour 2 or so; correct? 3 A. Yes. 4 Q. Can you show me in the article 5 where it's represented an analysis of the pool 6 data. You were telling me earlier that you 7 believe that there was a representation of an 8 analysis of pool data as well as of 9 stratification data? 10 A. Yes. 11 Q. Can you show me where the 12 representation of pool data is? 13 A. There's a number of 14 representations, I can show you one example of 15 that. 16 Q. Okay. 17 A. And that would be -- there's no 18 page number here. 19 Q. I know, I apologize for the 20 copy. If you can tell me how many pages into the -- 21 A. Fourth page, under the section 22 entitled suicidal acts. 23 Q. Up at the top it says suicidal 24 acts and ideations? Page 201 1 A. Yes. 2 Q. And then there's a table at the 3 bottom of the page? 4 A. Yes. The next section is 5 entitled suicidal acts. The second paragraph 6 presents the stratified analysis of that section, 7 the third paragraph presents the pooled 8 incidents. 9 Q. And the second paragraph under 10 suicidal acts where it says figure one and 11 presents the incidence differences, it refers to 12 figure one. Would that be Table Number 1? 13 A. No, that refers to figure one 14 on the next page. 15 Q. And, so, figure one is the 16 table representing the stratification? 17 A. Figure one shows each of the 18 trials included in the particular analysis, and 19 then the pooled line represents the incidence 20 difference, taking into account the 21 stratification. 22 Q. Okay. If we can go back to 23 Exhibit 4. In the fourth paragraph down, you say 24 we went over your comments, comma, my comments Page 202 1 with Vin and Mary today, that would be Vin Ramsey 2 and Mary Saylor? 3 A. Yes. 4 Q. The end result is a paper which 5 beefs up the inferences by stating, quote, 6 marginal significance, unquote, when P is less 7 than or equal to, is that what that's supposed to 8 mean? 9 A. Yes. 10 Q. Point one zero zero, period, 11 right? 12 A. Yes. 13 Q. What is -- what do you mean 14 when you say beefs up the inferences? 15 A. I'm not sure exactly without 16 having those comments that we addressed in front 17 of me, I'm not sure which part of those comments 18 I was referring to in particular. 19 Q. Can you give me an idea what 20 you meant? 21 A. The only thing I can recollect 22 is just what marginal significance means, when P 23 is less than .100. 24 Q. Okay. What does that mean? Page 203 1 A. The level of significance that 2 we've stated up front is the level .05. But 3 again, it's all relative, and, so, traditionally 4 statisticians have used the term marginal 5 significance when you get a P value less 6 than .100. And in this analysis, as described in 7 this paper, there are some, I believe, and I have 8 to look through here, but I believe there are 9 some trends which were marginally significant. 10 Q. What's the difference between 11 statistical significance and marginally 12 significant? 13 A. The only difference is that 14 statistical significance generally means at the 15 .05 level, P less than .05. Marginally 16 statistically significant means P less than .10, 17 so it's a relative term. 18 Q. Let me make sure I understand 19 it. The significance would be less with the .05? 20 A. You are more certain that 21 there's a treatment effect when your P value is 22 less than .05 than if it was just less than .10. 23 The smaller the P value is, the more certain you 24 are of something going on that can't be explained Page 204 1 by chance. So the effect is stronger when the P 2 value gets smaller. 3 Q. One in twenty as opposed to one 4 in ten? 5 A. Yes. 6 Q. Why is one in twenty -- why are 7 you more confident with one in twenty than you 8 are with one in ten? 9 A. That is signifying that you're 10 less likely to have a chance treatment difference 11 going on, that you're more likely to have the 12 real, true effect going on. So as the P value 13 gets smaller and smaller, that's telling you that 14 you're moving further and further away from this 15 null hypothesis that I mentioned yesterday which 16 is explaining chance and saying there's something 17 beyond chance that's leading to these results. 18 Q. You are, in effect, by using 19 the one in twenty, cutting the numbers in half, 20 aren't you? 21 A. I don't know what you mean 22 cutting the numbers in half. 23 Q. If you're starting with -- one 24 in ten is a lot more than one in twenty, right? Page 205 1 A. Yes, but remember from 2 yesterday when we were talking about the type one 3 error rate, which is the probability of saying 4 there is a treatment difference when there really 5 isn't. 6 Q. But that's a probability that 7 you insert within your statistical analysis, 8 right? 9 A. That's a prespecified level 10 before you do the analysis. See you set that 11 before you do the analysis and then you check to 12 see where your P value lies. So the smaller your 13 P value gets, the less likely you are of making 14 that type one error. 15 Q. But you're setting up -- when 16 you say you insert that, that's part of the 17 analysis that you decided to run on the data 18 right from the start, right? 19 A. Yes. We use the traditional 20 levels. .05 means statistical significance, and 21 marginal significance is .10. 22 Q. I guess what I'm trying to 23 understand is when did you make a decision to use 24 the .05, was it before the data was analyzed at Page 206 1 all or was that after some basic analysis of data 2 was done? 3 A. That's before analysis. That's -- 4 again, by tradition, that's what the FDA accepts 5 and that's what medical journals accept. And 6 oftentimes it's practiced to use marginal 7 significance of .10, but .05 is the gold 8 standard. 9 Q. What did you mean then in that 10 fourth paragraph where you say this is as far as 11 Charles is willing to go with the paper? 12 A. I believe that that statement 13 refers to Charles Beasley not desiring to put P 14 values in from the lumped or pooled data. 15 Q. Okay. He was willing to use P 16 values on stratified data but not on the pooled 17 data? 18 A. Yes, at this point in time. 19 Q. To your knowledge are there 20 other double-blind controlled studies that were 21 conducted in the United States that weren't taken 22 into consideration in the Beasley paper? I'm 23 talking about depression too, not other 24 indications. Page 207 1 A. I cannot recall. 2 Q. And again, your understanding 3 that part of the reason why the international 4 studies were not used in the reanalysis or the 5 analysis reflected in the Beasley paper is 6 because there were some differences between the 7 way that trials were constructed and the patients 8 that were admitted and things of that nature; 9 correct? 10 A. That's possible, yes. But 11 again, that's a medical determination. 12 Q. It's my understanding that 13 there were forty-six to forty-eight international 14 studies that were analyzed for presentation to 15 the advisory committee; is that correct? 16 A. I do not know the exact number 17 of those studies. 18 Q. But there was a narrowed down 19 number or a certain type of study that were 20 pooled similar to the seventeen that were pooled 21 for the Beasley study; correct? 22 A. It's my recollection there were 23 international studies that were analyzed, yes. 24 Q. Do you have any idea how many Page 208 1 studies were run on Fluoxetine across the world? 2 MR. MYERS: For what indication? 3 MS. ZETTLER: For any indication 4 whatsoever we'll start with. 5 A. No, I do not. 6 Q. Do you have any idea how many 7 studies were run on Fluoxetine for use with 8 depression only? 9 A. Around the world? 10 Q. Right. 11 A. No. 12 Q. Do you know how many were run 13 in the United States? 14 A. No, I don't know the exact 15 number. 16 Q. Did you ever question that they 17 were narrowing down the number of studies that 18 were going to be analyzed with regards to 19 suicidality? 20 MR. MYERS: Object to the form, what do 21 you mean by narrowing, narrowing down from what? 22 Q. Is it your understanding that 23 there were more clinical trials that were run 24 other than the seventeen in the United States? Page 209 1 A. No, I do not know. 2 Q. So I take it you never 3 questioned that there was a narrowing down of the 4 number of studies that were analyzed? 5 A. No. I personally, as a 6 reviewer and not an analyst in this particular 7 situation, questioned my staff to make sure they 8 felt good that these were the studies that needed 9 to be there, and they affirmed that, I guess, 10 they were the right studies there. 11 Q. Did anybody ever say to you I 12 think we should include this study or this study 13 in this analysis? 14 A. No. 15 Q. Were they privy to all the 16 studies that were done on depression in the 17 United States? 18 MR. MYERS: They who? 19 MS. ZETTLER: The staff. 20 A. I don't know. 21 Q. What do you mean in the 22 beginning of the third paragraph in Exhibit 4 23 when you say, thus, anything we do now to make 24 lumping more attractive is not productive? Page 210 1 A. As I stated before, we believe 2 the stratified analysis to be the primary 3 analysis of the paper, and by this statement, I'm 4 just saying we -- from a statistical standpoint, 5 we don't want to make that the predominant or 6 primarily analysis, we want to stick with the 7 stratified analysis up front as the primary 8 analysis. 9 Q. Right above that you say, in 10 the paragraph above that, before we had always 11 heard the studies are the same, go ahead and 12 pool, my belief in their similarity has dropped. 13 Who told you that the studies were the same? 14 A. I don't remember. 15 Q. Was it Beasley? 16 A. I don't want to ascribe it 17 totally to Beasley. I know that as a physician 18 that works with the statistical group, he 19 certainly was involved in those kinds of 20 discussions. 21 Q. He was a lead author on this 22 paper; correct? 23 A. Yes. 24 Q. He's the one who went and Page 211 1 rereviewed the seventeen trials and found that 2 there were differences; correct? 3 MR. MYERS: Differences in what? 4 Q. Differences in the trials, 5 according to what you say in this letter or this 6 memo, I'm sorry. 7 A. Yes, and he knew there were 8 differences, it's not he rereviewed it and found 9 differences per se. We knew trials were 10 different up front, as we said before. The point 11 is, how different are they, are they different 12 enough or not different enough where you can look 13 at them together. So up front, a priorty, 14 everyone knows there's differences between 15 trials. 16 Q. But prior to this memorandum, 17 somebody, be it Beasley or somebody else working 18 on this project, represented to you and your 19 group that the information that was contained or 20 the study design, things of that nature, were 21 similar enough that you could pool the data and 22 analyze it that way; correct? 23 A. No. The trials were similar 24 enough where you could do analysis, whether it be Page 212 1 stratified or pooled, that's not the issue. The 2 trials were similar enough where you could use 3 them and analyze them together, somehow. And 4 then as Charles dug in, from my understanding a 5 little bit more, he even saw that there were more 6 differences between the trials, but not to the 7 point where you would say we can't do these same 8 analyses on these trials. 9 Q. But that's not my question. My 10 question is: Prior to this memo, it was 11 represented to you and your staff that there was 12 enough similarity between these seventeen trials 13 that data could be pooled to be analyzed, per 14 this memorandum. 15 A. Yes. 16 Q. Did anybody on your staff or 17 you yourself ever review the seventeen trials to 18 make sure that the variables, so to speak, were 19 similar enough, the variations in the different 20 studies? 21 A. Yes. 22 Q. Who did that, did you? 23 A. I reviewed trial demographics, 24 such as those suggested in the appendix in this Page 213 1 paper which shows different demographic 2 cross-studies. 3 Q. And those demographics were 4 provided to you by the medical department; 5 correct? 6 A. I'm not sure who produced 7 those. 8 Q. You didn't go and search the 9 files for demographic data, you were provided -- 10 A. No, I personally did not. 11 Q. Let me finish my question. 12 A. I'm sorry. 13 Q. You were provided with that 14 information, you didn't do a review of each of 15 those clinical trial data bases on your own; 16 correct? 17 A. That's correct. 18 Q. Yesterday we were talking about 19 your analysis is only as good as the data that 20 you're given to work with; correct? 21 A. Yes. 22 Q. It's kind of like the garbage 23 in, garbage out theory, right? 24 A. Yes. Or clean data in, clean Page 214 1 data out. 2 Q. That depends on your point of 3 view. Do you have any understanding as to the 4 Hamilton depression rating scale other than the 5 fact that they use question number three to rate 6 suicidality? 7 A. What kind of understanding are 8 you -- 9 Q. Like the validity of the scale, 10 are you aware of whether or not the scale is a 11 valid determination of suicidality? 12 A. To the best of my knowledge, 13 there have been studies done showing that the 14 Hamilton depression scale is valid in measuring 15 depression, and it has been the primary variable 16 that has been used in many clinical trials. 17 Q. What about suicidality, is it a 18 valid measure of suicidality? 19 MR. MYERS: Are you asking him from a 20 statistical standpoint? 21 MS. ZETTLER: To the best of his 22 knowledge. 23 Q. Do you consider, as a 24 statistician, using merely question three of the Page 215 1 Hamilton rating scale a valid measure of 2 suicidality? 3 A. As a statistician, from a 4 numerical standpoint, I consider it to be an 5 accurate and valid reflection -- 6 Q. Okay. 7 A. -- of suicidality. But I 8 cannot speak to the medical process behind that. 9 Q. Okay. You talked earlier about 10 working -- your having worked at some point with 11 Dr. Ivan Miller on a proposed rechallenge study; 12 correct? 13 A. On a proposed validation -- 14 Q. Validation study of his scale, 15 his suicidality scale; correct? 16 A. Yes. 17 Q. Did you ever look at that 18 scale? 19 A. I believe I may have. 20 Q. Okay. Are you aware that it's 21 more than one question long? 22 A. Yes. 23 Q. Okay. In fact it's quite a few 24 questions long, isn't it? Page 216 1 A. I'm not certain of the exact 2 number, but I know it's more than one question 3 long. 4 Q. And are you familiar with what 5 the question three from the Ham-D scale is, what 6 the actual question is? 7 A. I can't recall the actual 8 question. 9 Q. Okay. But you have no problem 10 with using simply one question from a depression 11 scale to analyze the incidence of suicidality in 12 a large number of people; correct, from a 13 statistical point of view? 14 A. As a statistician, I have no 15 problem whatsoever. 16 Q. If it were to be shown that 17 using merely one question from the Hamilton 18 depression rating scale to measure suicidality 19 was an invalid way or inadequate way to measure 20 suicidality, what would that do to your analysis 21 of the data? 22 A. I can't conceive that invalid 23 or inadequate are the same thing there, so I'm 24 not sure how to address your question. Page 217 1 Q. Let's start with inadequate. 2 A. As a statistician, I believe 3 the Hamilton question item three to numerically 4 be valid. We have -- validity is measured in a 5 number of ways, and we have seen changes, for 6 example, over time, that's one way to validate a 7 scale. We saw changes even in this study. 8 Q. All you're doing is validating 9 the fact that their scores on that particular 10 question changed; correct? 11 A. Numerically, that's all I can 12 speak to. 13 Q. So you don't know if a change 14 in the Hamilton depression rating scale question 15 number three regarding suicidality really is a 16 determination of whether or not a person has 17 become more suicidal, do you? 18 A. As a statistician, no, that's 19 not my call to make. 20 Q. So you're relying completely on 21 the representations of people at Lilly that this 22 is a reliable determination of suicidality as far 23 as the actual ability of that question to 24 determine suicidality; correct? Page 218 1 A. I'm primarily relying upon the 2 medical colleagues at Lilly in making those 3 determinations. 4 Q. You never did any independent 5 research as to the validity of question three of 6 the Hamilton depression rating scale to gauge 7 suicidality, did you? 8 MR. MYERS: Medically? 9 MS. ZETTLER: Yes. 10 A. Not medically. I could never 11 do that now or in the future. 12 Q. You can't go to the library and 13 look up and do research on whether or not any 14 literature has been published as to whether or 15 not that is a valid determination of suicidality? 16 A. Yes, I can, as a statistician. 17 In fact I have looked at some original papers 18 where the author of the Hamilton depression scale 19 did factor analyses and things like that to look 20 at validity there. I don't recall the details of 21 those papers, but I have taken the opportunity to 22 look at that. 23 Q. Did the author of the Hamilton 24 depression rating scale say that question number Page 219 1 three was a valid rating of suicidality? 2 A. I don't recall. 3 Q. If that was the case, that 4 question number three is a valid rating of 5 suicidality, why weren't they going to use that 6 in the rechallenge protocol as opposed to 7 rewriting the entire scale to use for the 8 protocol? 9 A. My perception is that the 10 medical experts believed one, that the Hamilton 11 item three is a valid way of measuring 12 suicidality, but they wanted to look at other 13 aspects of suicidality that that particular item 14 did not encompass. So they wanted to go beyond 15 that. But they did not believe that it was an 16 invalid, that Hamilton item three was invalid. 17 Q. What if the doctors at Eli 18 Lilly, one of the research physicians, were to 19 say that the Hamilton depression rating scale 20 question three was not a very good indication of 21 suicidality, would that affect your confidence in 22 your ability to show that Fluoxetine does not 23 cause suicidality in any way? 24 A. As a statistician, numerically Page 220 1 I could still accept that as being valid and 2 accurate numerically. Now the clinical meaning 3 behind that question is a different thing, and I 4 have to defer to clinicians on that. But from a 5 numerical and accuracy standpoint, it does not 6 diminish my confidence in the actual analysis per 7 se, it might change the meaning behind that 8 analysis. 9 Q. You testified earlier that you 10 believed that your analysis of the seventeen 11 clinical trials helped prove in part that 12 suicidality was not more likely to occur with 13 Fluoxetine as to other drugs; correct? 14 A. Our analysis did not 15 demonstrate an association of increased risk with 16 Fluoxetine. 17 Q. It's your belief that 18 Fluoxetine does not cause suicidality in patients 19 any more frequently than it does in placebo or 20 tricyclics or other comparitors; correct? 21 MR. MYERS: Before he answers, let me 22 object to the form because you're assuming cause 23 and effect in all three treatment groups, and 24 there's been no testimony about that. Page 221 1 Q. Is that your belief? 2 A. Would you restate the question? 3 Q. Sure. 4 MS. ZETTLER: Why don't you go ahead 5 and read it back. 6 (THE COURT REPORTER READ BACK THE 7 REQUESTED TESTIMONY.) 8 A. Yes. 9 Q. And your opinion is based in 10 part on the data that was presented to you to 11 analyze, isn't it? 12 A. Yes. 13 Q. And your opinion is based in 14 part on the validity of that data; correct? 15 A. Yes. 16 Q. If it was shown that the 17 validity of data that you used, that you were 18 given to analyze was suspect, would that change 19 your opinion at all? 20 MR. MYERS: Object to the form, what do 21 you mean by suspect? 22 MS. ZETTLER: It wasn't accurate, it 23 was invalid, it was worthless. 24 A. I can't answer that fully Page 222 1 because there's quite a range of possibilities 2 that that question leads to. 3 Q. Why don't you give me some 4 possibilities. 5 A. Those are medical 6 possibilities, I can't even begin to outline them 7 because again, it would come from a clinician and 8 the interview process and what they're picking up 9 and measuring with that particular item. 10 Q. So you're confident in your 11 opinion, and your opinion is based solely on the 12 validity of the data that's presented to you by 13 the medical personnel at Lilly; correct? 14 A. Yes. 15 Q. Have you ever questioned the 16 validity of that information? 17 A. I can't answer that directly 18 with a yes or no, I can tell you that, for 19 example, I have gone back and done some 20 literature review just to learn about the Ham-D 21 and that kind of thing. So question, yes, as a 22 statistician, I'm a reasonable skeptic and I 23 would want to find out what's behind the data. 24 Q. You testified earlier that it Page 223 1 was your belief that the Lilly doctors, I think 2 you said, cast a broader net to try to determine 3 whether or not certain incidents were suicide 4 related, things of that nature; correct? 5 A. Yes. 6 Q. Were you ever involved in 7 reviewing those individual cases? 8 A. No. 9 Q. Were you ever involved -- were 10 you ever in a meeting in which those individual 11 cases were reviewed? 12 A. I don't remember. 13 Q. Do you have any knowledge as to 14 the original number of cases that there were that 15 were reviewed? 16 A. No. 17 Q. Did you ever question the 18 number of cases that were reviewed? 19 MR. MYERS: Question in what way? 20 MS. ZETTLER: Any way. 21 A. I may have asked about the 22 algorithms they were using, and just checked that 23 up with them. But as far as questioning, saying, 24 you know, are you sure it's this number and not Page 224 1 this number, I don't remember doing that. 2 Q. Did you ever -- have you ever 3 seen the differences between the original number 4 and the number that was analyzed for purposes of 5 the paper? 6 A. I can't recall that number, I 7 don't believe so. 8 Q. Is that something that you 9 could do now if you wanted to, could you go back 10 and say could I see the original number of 11 alleged suicidal acts and compare them to the 12 number that were eventually given to me to 13 analyze? 14 A. I could certainly ask the 15 question, I don't know if that's the right 16 question to ask, but I could ask that question. 17 Q. Could you say to somebody I 18 would like to see the original number of 19 suspected suicidal acts or suicidal ideations, 20 could you say that, could you ask them that? 21 A. I could ask them that. 22 Q. Could they provide you with 23 that information? 24 A. Yes, I am pretty sure they Page 225 1 could. 2 Q. Then you could go back actually 3 to the article and compare that with the number 4 of alleged suicidal acts or incidents of a 5 suicidal ideation and then compare the numbers, 6 right? 7 A. Yes. 8 Q. Okay. So you have at your 9 disposal both of those lists or both of those 10 numbers; correct? 11 A. I personally don't have them at 12 my disposal, but I could ask these questions. 13 Q. And you could theoretically get 14 that information, right? 15 A. I believe so. 16 Q. You've never done that? 17 A. No. 18 Q. Has anybody on your staff ever 19 done that? 20 A. I'm not sure. 21 Q. Did you have any say in what 22 was to be considered a suicidal act or suicide 23 act and what was not to be considered a suicide 24 act? Page 226 1 A. No. 2 Q. Your job was simply to take the 3 numbers that they presented to you and do an 4 analysis from those numbers; correct? 5 A. To clarify, I did not do these 6 analyses, so I didn't take these numbers, I 7 reviewed analyses where numbers were in the data 8 base and that data base was analyzed. 9 Q. So your department's job was to 10 take the numbers that were presented to you by 11 the clinical research physicians at Lilly and do 12 the analysis from those numbers? 13 A. To clarify, our department's 14 job was to analyze the data base that was given 15 to us. So we weren't given numbers, per se, we 16 were given a data base, and we analyzed that data 17 base. 18 Q. And the data base consisted of 19 a finite number of people? 20 A. Yes. 21 Q. It wasn't the entire number of 22 people who have ever been given Fluoxetine in 23 clinical trials to that point; correct? 24 A. I'm not sure. Page 227 1 Q. Who would know that? 2 A. I'm not sure exactly who would 3 know that. 4 Q. Can you give me an idea of who 5 might know that? 6 A. It would be someone in the 7 clinical data base management group. 8 Q. Systems, somebody in systems? 9 A. Yes. 10 Q. The number of patients that 11 were analyzed for purposes of the Beasley article 12 were approximately, I believe, twenty-five 13 hundred, wasn't it something along those lines? 14 A. No, it was not twenty-five 15 hundred. If you add up the three numbers under 16 the design on the abstract, I believe it comes to 17 thirty sixty-five. 18 Q. Okay, thirty sixty-five. I'm 19 not real good at adding things in my head. I'm 20 not really good at adding things on paper, 21 either. To your knowledge were there more than 22 three thousand sixty-five patients who had been 23 exposed to Fluoxetine prior to that date? 24 MR. MYERS: In the whole world or in Page 228 1 clinical trials? 2 MS. ZETTLER: In the clinical trials. 3 Q. Let me ask it this way: If 4 there was a cut-off date that was applied to the 5 data bases, in other words there was a certain 6 point in time where they decided we're not going 7 to concern ourselves with information that has 8 come in to the data bases through clinical trials 9 past a certain date; correct? 10 A. Yes. 11 Q. Do you remember what that date 12 was? 13 A. No, I do not. 14 Q. So we'll use that date as -- 15 when I say the cut-off date, we'll just use that 16 date, okay, and I'm going to ask you general 17 questions and I'll try not to ask you anything 18 too specific about it, because frankly, I 19 probably can't anyway. But up to that cut-off 20 date, do you have any knowledge as to whether or 21 not more than three thousand sixty-five people 22 had been exposed to Fluoxetine through clinical 23 trials? 24 MR. MYERS: Excuse me. In these Page 229 1 seventeen trials or in any clinical trials? 2 MS. ZETTLER: No, any clinical trials. 3 A. I believe there were 4 international trials, for example, that weren't 5 included in that thirty sixty-five. 6 Q. That were included or weren't? 7 A. Weren't. 8 Q. And there were -- do you know 9 if there were other clinical trials that had been 10 conducted in the United States as well, other 11 than the seventeen? 12 A. I'm not sure. 13 MS. ZETTLER: Can you go back about 14 three questions? 15 (THE COURT REPORTER READ BACK THE 16 REQUESTED TESTIMONY.) 17 Q. The data base that your 18 department analyzed, do you know if that 19 contained information on every patient that had 20 been exposed to Fluoxetine through clinical 21 trials up to that point or if it was narrowed 22 down to the three thousand sixty-five people? 23 A. I'm not sure exactly what the 24 total number was. Page 230 1 Q. Have you done similar analyses, 2 not meaning suicidality, where you have taken, 3 for instance, information from one clinical trial 4 and analyzed it off the data base? 5 A. Have I personally analyzed one 6 clinical trial in general? 7 Q. Right. 8 A. Yes. 9 Q. Have there been people excluded 10 from analysis on those trials? 11 MR. MYERS: Which trials? 12 Q. We don't need to be this 13 specific, I'm just talking generally in the 14 analyses that you've done similar to that on a 15 clinical trial, have people been excluded from 16 analysis? 17 A. The general approach that our 18 group takes in analyzing data is what is known as 19 the intent-to-treat approach, everyone randomized 20 gets analyzed. 21 Q. But if they're not randomized, 22 they're not analyzed? 23 A. That's correct. 24 Q. Is there a mechanism on SAS or Page 231 1 some other mechanism that you have to go through 2 physically to determine which people have been 3 randomized and which have not? 4 A. No. 5 Q. Is that something that the 6 computer will do for you? 7 A. That's something that the 8 systems group traditionally has done for us. The 9 statisticians get a data base of randomized 10 patients because our analysis is intent to treat. 11 Q. So to your knowledge was 12 something similar done with the three thousand 13 sixty-five patients, in other words could there 14 ever have been a larger group of patients and 15 then they were narrowed down to three thousand 16 sixty-five? 17 A. I'm not sure. Our traditional 18 approach is if you're randomized, you're 19 analyzed. But again, I didn't analyze this data 20 so I'm not one hundred percent sure. 21 Q. So if this was done under the 22 traditional approach, your staff would have just 23 been given the data on the three thousand 24 sixty-five people? Page 232 1 MR. MYERS: That were randomized? 2 MS. ZETTLER: The three thousand 3 sixty-five people that were analyzed. 4 MR. MYERS: But he's talking about 5 intent to treat randomized. 6 Q. So are you saying that people 7 that were not randomized were taken into 8 consideration for the Beasley article? 9 A. This article only deals with 10 randomized patients. 11 Q. Okay. So my question is: 12 Would you have been given a data base with more 13 than just these randomized three thousand 14 sixty-five patients, and you would have to go 15 through and extract just the randomized patients 16 who were doing it the traditional method? 17 A. I can't state with certainty 18 because I didn't analyze the data, but I do 19 believe they were given a data base with the 20 randomized patients. 21 Q. And assuming the three thousand 22 sixty-five were the only patients that were 23 randomized, it would just be those patients? 24 A. Correct. Page 233 1 Q. When that data base was 2 presented to your staff, would it already have 3 had a determination made on whether or not an 4 individual case was suicide related? 5 A. Yes. 6 Q. To your knowledge did somebody 7 from the medical department such as someone like 8 Doctor Beasley or one of the other clinical 9 research physicians, previous to transferring 10 that data base for your analysis, review those 11 individual cases? 12 A. What do you mean by review? 13 Q. Look at the individual cases 14 for suicidality to make a determination as to 15 whether or not it was actually a suicidal -- 16 suicide attempt, for instance? 17 A. Yes, all cases were reviewed. 18 Q. And prior to that, the cases 19 had already been reviewed as part of the 20 collection of the information for clinical report 21 forms, isn't that true, prior to Doctor Beasley's 22 analysis or Doctor Beasley's article being 23 written? 24 A. To the best of my knowledge, Page 234 1 this algorithm was used to review the case report 2 forms for possible suicidality, and that 3 determination was then included in the data base. 4 Q. Okay. I'm talking about prior 5 to that analysis, that review, I'm talking about 6 in the normal clinical trial process. 7 A. While the studies were ongoing? 8 Q. Right. Somebody reviews that 9 information at the time to see if it's, for 10 instance, a suicide attempt or something along 11 those lines; correct? 12 A. I believe so. Again, I'm in 13 the statistical department, and the clinical 14 research physician's job is to review the data 15 and monitor the study as it progresses. 16 Q. What's been your experience 17 when you've done the analysis on a clinical 18 trial, is it your experience that somebody has 19 reviewed the case report forms for various 20 purposes, not just for adverse events but for 21 various purposes; correct? 22 A. Yes. 23 Q. And that is so that the data 24 can come to your department in as pristine a form Page 235 1 as possible; correct? 2 A. Yes. 3 Q. That analysis is done just in 4 the normal course of analyzing and preparing the 5 information gleaned from a clinical trial; 6 correct? 7 A. Yes. 8 Q. Is it your experience that 9 prior to your analysis, a reanalysis analysis of 10 information, such as that done by Doctor Beasley 11 and his people with regards to the BMJ article, 12 was done as a regular course? 13 A. I'll need some help clarifying 14 this word reanalysis, I'm not sure what you mean 15 by that. 16 Q. We've already established that 17 in your experience as part of the clinical trial 18 process somebody at Lilly reviews the information 19 gleaned from the clinical trial as it comes in; 20 correct? 21 A. Yes. 22 Q. And eventually you or somebody 23 in your department gets this information in the 24 data base to analyze statistically; correct? Page 236 1 A. Yes. 2 Q. I want to find out if between 3 the initial review that's done by somebody at 4 Lilly when the trial information comes in and the 5 time you guys get it over in statistics, if it's 6 normal for somebody to rereview the quality or 7 the accuracy of the information before it's sent 8 to you or something. In other words, what Doctor 9 Beasley did in preparation for the BMJ article, 10 is that something that is done as a normal matter 11 of course? 12 A. I do not know. 13 Q. Has it been your experience on 14 any of the other clinical trials or any clinical 15 trials that you have worked on that that is done? 16 A. I'm not familiar with the 17 medical quality assurance process. 18 Q. Okay. 19 A. They may review, make sure it's 20 clean, and that might take a long time or short 21 time, I'm not sure how many times they actually 22 review the data. 23 Q. Okay. 24 MS. ZETTLER: Can we take a quick Page 237 1 break? 2 MR. MYERS: Sure. 3 (A SHORT RECESS WAS TAKEN.) 4 Q. (BY MS. ZETTLER) Let's talk a 5 little bit about how the data was -- not 6 analyzed, but once you had the data base, what 7 you were able to do with it, what were the 8 mechanisms to find out what was a suicidal act 9 and things of that nature. I believe in the 10 article it said that there were some event terms 11 that were used, were included, such as overdose, 12 things of that nature? 13 A. On the second page of the 14 article under methods, the third paragraph talks 15 about potential indication of suicidal acts were 16 first identified electronically, et cetera. 17 Q. The data for adverse events was 18 collected by people on your staff by looking at 19 different event terms; correct, such as -- 20 A. Did you say our staff collected 21 the data? 22 Q. No, was looked at. If you -- 23 if somebody on your staff wanted to take that 24 data base and find out how many people suffered Page 238 1 from what was considered to be a suicidal act, 2 okay, you would look under -- you would pull up a 3 certain event term, right? 4 MR. MYERS: When you say you, who are 5 you talking about? 6 Q. Generally within your 7 department. 8 A. My staff would get a data base 9 where a suicidal act, for example, would be a 10 variable that was labeled such, and they would 11 just -- 12 Q. So they were already labeled a 13 suicidal act? 14 A. That's not the term in the data 15 base, but it would have some name like that. 16 Q. What was the term in the data 17 base? 18 A. I don't know. 19 Q. Was that an adverse event term 20 that was used or was it a label? In other words, 21 did the information come to your department 22 labeled suicidal ideation, suicide attempt or 23 some way you could already -- you didn't have to 24 go through and categorize each of the events Page 239 1 yourself? 2 A. Our department does not 3 categorize anything ourselves, we take what has 4 already been categorized and we analyze it. 5 Q. What I'm trying to find out, 6 because I don't know, okay, so it's going to be a 7 little bit choppy, is if you ran specific event 8 terms -- do you know what an event term is, an 9 adverse event term? 10 A. Yes. 11 Q. Are you familiar with ELECT, 12 the ELECT dictionary? 13 A. I know the dictionary. 14 Q. Okay. Were -- was that 15 information categorized per ELECT term? 16 A. I don't know. I can only tell 17 you what I do know about the data base that was 18 given to my staff, and that is, for example, once 19 the medical group had decided on which patients 20 were to be categorized as having a suicidal act 21 or not, that would fall under a certain variable, 22 some name given to it, like suicidal act or 23 whatever, and then that variable would be 24 analyzed. So our staff does not go into certain Page 240 1 ELECT terms and pull things together, it just 2 takes that variable that's on this data base 3 that's been given to our staff, and focus on that 4 one column or one variable and then analyze it. 5 Q. So that variable would come, 6 just using, for example, suicide attempt, and it 7 would say sixteen. Or would it be broken down 8 even further, in other words were you given a 9 one-lump number or were you given individual case 10 numbers from each study? 11 A. Individual patients are 12 represented in that data base. For example, 13 under that variable it would say yes or no or 14 zero one or some representation of whether that 15 patient had an act or not. 16 Q. So the determination as to 17 whether or not that was an actual act was already 18 made when you have -- in your data base? 19 A. Yes. 20 Q. And that determination was made 21 by people in medical as far as you know? 22 A. Yes. 23 Q. Does that data base contain a 24 representation as to possible suicide acts that Page 241 1 were determined not to actually be suicide acts? 2 A. I'm not sure. 3 Q. Again, it's your belief that at 4 least initially a broad net was cast to try to 5 determine anything that could have been a suicide 6 act or suicidal ideation; correct? 7 A. Yes. 8 (PLAINTIFFS' EXHIBIT NO. 6 WAS 9 MARKED FOR IDENTIFICATION AND 10 RECEIVED IN EVIDENCE.) 11 Q. Have you had a chance to look 12 at Exhibit Number 6? 13 A. Yes. 14 Q. Are you familiar with this 15 exhibit, have you seen this exhibit before? 16 A. I may have. 17 Q. Have you seen similar 18 documents? 19 A. I'm not sure, this was produced 20 by someone in the medical group. It's produced 21 by someone not in statistics. 22 Q. But it was produced to us as 23 part of your file, individual file. 24 A. Okay, yes. Page 242 1 Q. Have you had a chance to look 2 at any of these summaries? 3 A. Just briefly here. 4 Q. Why don't you take a couple of 5 minutes and look at them all. 6 A. Okay. 7 (THE WITNESS COMPLIES.) 8 Q. Have you had a chance to look 9 at the exhibit? 10 A. Yes. 11 Q. Let's talk about the first one 12 listed on the second page of the exhibit, Pz 2441 13 1806, okay? 14 A. Yes. 15 Q. On the left-hand side, it says 16 patient identification. Is it BPHC two five, and 17 then there's something blanked out. Do you know 18 what the BP stands for? 19 A. I'm not certain. 20 Q. Okay. Do you have any idea? 21 A. This is an identification for a 22 particular project, I believe, but I'm not sure 23 what BP means or how it refers to a project. 24 Q. Have you seen a PIP number? Page 243 1 A. Yes. 2 Q. Is this a PIP number here? 3 A. This was a few years ago when I 4 analyzed data. It might be, yes. 5 Q. PIP stands for project 6 investigator patient, doesn't it? 7 A. Yes. 8 Q. Does it look like a project 9 number to you on the left-hand side? 10 A. Yes. 11 Q. Isn't a word project given four 12 initials generally? 13 A. My understanding of PIP is 14 right, the project is four. 15 Q. Four initials? 16 A. Four letters. 17 Q. The investigator is a number 18 and the patient is a number or initials, 19 depending on when -- generally it was numbers, 20 right, for patients? 21 A. Yes, the PIPs I have been 22 involved with were numbers for patients. 23 Q. Have you ever seen a study 24 initial just containing two initials as opposed Page 244 1 to four? 2 A. No, because I analyzed U.S. 3 studies. 4 Q. Okay. 5 A. This is international clinical 6 trials here, so I'm not sure -- I'm not familiar 7 with the international clinical trials. 8 Q. So the international -- could 9 the P on the left-hand side be a country code? 10 A. It might be. 11 Q. Okay. Regardless, the second 12 to the last column on the right-hand side says 13 description of possible event; correct? 14 A. Yes. 15 Q. Okay. And under that for the 16 first patient it says two suicide attempts on 17 March 9th, '87 with twenty to twenty-five tablets 18 of mogadon done, and on March 10th, '87 with 19 twenty tablets of unknown anti-depressant; 20 correct? 21 A. Yes. 22 Q. Then it has a category right 23 next to that, reason for not meeting case 24 definition, study drug was discontinued eight Page 245 1 days prior to first overdose, discontinued due to 2 severe rash; correct? 3 A. Yes. 4 Q. Is it your understanding that 5 this person was not analyzed as somebody who had 6 suffered a suicideal act? 7 MR. MYERS: Before he answers, let me 8 object to the form. When you say not analyzed, 9 for what, because we've been talking about the 10 Beasley article. 11 MS. ZETTLER: For the Beasley article -- 12 I'm sorry, I take that back, for the advisory 13 committee meeting. 14 A. Just based on the titles of 15 this table one, it's my impression that these 16 patients did not meet the case definition of 17 suicidal act and hence were not included in a 18 particular analysis for the -- of the 19 international data for the advisory committee 20 meeting. 21 Q. So in other words, patients 22 such as this that were found to not meet the case 23 definition would not have been included in the 24 analysis as people who had suffered suicidal acts Page 246 1 or suicidal ideation? 2 A. Not necessarily because I'm not 3 familiar with all the analyses that were done on 4 that data. But these patients were not included 5 in a particular analysis of international data. 6 Q. Would patients -- without 7 worrying about whether or not these people are in 8 international trials or getting to anybody 9 specific, but generally would these types of 10 patients who were determined not to fall under 11 case definitions have been included in the 12 analysis for the Beasley BMJ article? 13 MR. MYERS: Let me object to the form. 14 Are you saying was the criteria for the Beasley 15 paper the same as whatever criteria is reflected 16 in this paper, in this document? 17 MS. ZETTLER: Yes, I guess that's a 18 good question, thanks. 19 Q. Was the criteria the same 20 across the board or were different criteria used 21 for different analyses? 22 A. I'm not certain, I don't know. 23 Q. Let me ask you this generally, 24 okay, if a patient was found, for the purposes of Page 247 1 the Beasley article, not to fit the case 2 definition used in the Beasley article of 3 suicidal act or suicidal ideation, if they were 4 found not to meet that criteria, were they then 5 not included in the analysis of cases of suicidal 6 acts or suicidal ideation? 7 A. That's not true. 8 Q. Okay. Tell me why it's not 9 true. 10 A. All patients were included in 11 the analysis, but those particular patients would 12 not be included as those who had suicidal acts or 13 ideation if they didn't meet the case definition. 14 But all patients were included in the analysis, 15 all three thousand sixty-five. 16 Q. Was there a differentiation 17 made in your analysis for people who had 18 initially been reported to possibly suffer a 19 suicidal act as opposed to people who were 20 determined to be a suicidal act or was it broken 21 down into these people suffered a suicidal act 22 and other people didn't? 23 MR. MYERS: Are you asking if that's 24 what the paper does? Page 248 1 MS. ZETTLER: Right. 2 A. I'm not exactly clear. I can 3 explain the data base that we got and tell you 4 how each patient was identified. 5 Q. Okay. 6 A. If that's what you're getting 7 at. 8 Q. Sure, let's start with that. 9 A. Using the case definition 10 that's articulated here in the Beasley paper, our 11 data base would have, for example, three thousand 12 sixty-five patients. For each patient under this 13 one particular variable name, there would be 14 either a zero one or yes or no, I'm not sure 15 exactly what was there, to tell the statistician 16 analyzing that data whether that person was 17 declared to have an act or not. And, so, for 18 each patient there was a yes or no or yes or no, 19 all the way through it. So all that data base 20 was analyzed, the yeses were counted up. 21 Q. What would happen with the no 22 people, would there be -- is there something in 23 that article that says that there were, just as 24 an example, three hundred patients that were Page 249 1 analyzed for possible suicide attempt and it was 2 determined that only, and I don't remember the 3 exact numbers of the papers, I'm just going to 4 use this as an example, ten people that were 5 determined to have suffered an actual suicide 6 attempt? 7 A. No. 8 Q. Just concentrated on those 9 people who were determined to have suffered an 10 actual suicide attempt? 11 A. Our staff was given the data 12 base that had been declared up front by the 13 physicians using the algorithm as to whether each 14 individual patient met the definition or not. If 15 the patients do not meet the definition, there 16 would not have been a yes in the data base. 17 Q. And that definition was created 18 by Lilly medical personnel? 19 A. I'm not exactly sure who all 20 was involved in the creation of that. As the 21 paper, BMJ paper, points out, there's a lot 22 involved in that definition, I'm not sure if it's 23 just Lilly or not. 24 Q. So it could have also been with Page 250 1 the outside consultants that were used? 2 A. Possibly. 3 Q. It could also have been people 4 in upper management at Lilly? 5 A. I'm not sure. 6 Q. But it was something created at 7 Lilly, the definition, to the best of your 8 knowledge? 9 A. What do you mean created at 10 Lilly? 11 Q. The definition was written at 12 Lilly. 13 A. Yes. 14 Q. Okay. 15 Q. And that goes for the 16 definition of suicide attempt as well as suicidal 17 ideation? 18 MR. MYERS: For the purposes of the 19 paper? 20 MS. ZETTLER: The Beasley article, 21 right. 22 A. Yes, in the definition section 23 of the BMJ paper, act is defined and ideation is 24 defined. Page 251 1 Q. Where is the definition 2 section? 3 A. That's page two. 4 Q. Okay. Definitions up in the 5 upper right-hand corner? 6 A. Yes. 7 Q. Says suicidal act is defined as 8 any behavior undertaken purposely for which the 9 outcome was likely to be self harm and where no 10 explicit data suggested that suicide had not been 11 intended; correct? 12 A. Explicit data, yes. 13 Q. You're just clarifying that it 14 had to be explicit data? 15 A. Yes. 16 Q. What is explicit data? 17 A. I'm not sure exactly what 18 explicit data is. 19 Q. Okay. And it goes on to say 20 actions that might be described as suicidal 21 gestures were not excluded. So they were 22 included in the group that fit the first portion 23 of the definition, the suicidal act, as defined 24 as any behavior undertaken purposely from which Page 252 1 the outcome was likely to be self harm; correct? 2 A. As I read this, yes. 3 MR. CLEMENTI: What exhibit are you 4 reading from? 5 MS. ZETTLER: Exhibit 5. 6 Q. It also goes on to say suicidal 7 act had to have occurred before or during the day 8 following the last day of double-blind treatment 9 in compliance with the trial protocol; correct? 10 A. Yes. 11 Q. This time limit was adopted for 12 three reasons, colon, post discontinuation data 13 had not been collected as part of the trial, 14 semicolon, the end of participation in the trial 15 or withdrawal of the study treatment, or both, 16 might have influenced an event occurring after 17 discontinuation, semicolon, and other drugs might 18 have been started after the end of the study 19 treatment period; correct? 20 A. Yes. 21 Q. Do you have any idea what the 22 half-life of Fluoxetine is? 23 MR. MYERS: I think that's a medical 24 question, but if he has independent Page 253 1 understanding, he can answer it. 2 MS. ZETTLER: I'm not asking him as a 3 doctor, I'm asking him personally if he has any 4 idea what the half-life of Fluoxetine is. 5 MR. MYERS: I object to the form, but 6 he can answer it if he knows. 7 A. I'm not for sure, I'm not 8 certain, I believe it's about a day. 9 Q. A day? 10 A. Twenty-four hours. That number -- 11 again, I'm not sure exactly what the half-life of 12 Fluoxetine is. 13 Q. Okay. Would it bother you if 14 the half-life of Fluoxetine was more like three 15 or four days? 16 MR. MYERS: I object to the form. 17 Q. As far as this definition is 18 concerned? 19 MR. MYERS: Which definition? 20 MS. ZETTLER: The definition I just 21 read where it talks about the 22 post-discontinuation data. 23 MR. MYERS: Well, I object to the form, 24 this is all subject to medical expertise and he's Page 254 1 not a medical expert. 2 Q. Would you have a problem with 3 limiting the time in which a suicidal act had 4 occurred per the definition set out in the 5 Beasley article if you knew that the half-life of 6 Fluoxetine was much longer than one day? 7 A. Not necessarily. 8 Q. Okay. Why not? 9 A. As is mentioned here, for 10 example there might be other drugs that are 11 started following discontinuation, and to include 12 that data possibly might influence how you 13 compare the treatment groups. The patients might 14 be taking different kinds of treatments after the 15 trial is over. 16 Q. It's true that Lilly became 17 aware, at least, of suicide attempts that 18 occurred longer than a day after discontinuation 19 of treatment; correct? 20 MR. MYERS: In what? 21 MS. ZETTLER: Either as a result of a 22 clinical report from a trial or from a 23 spontaneous event. 24 MR. MYERS: At what point in time? Page 255 1 MS. ZETTLER: Throughout the clinical 2 trials, the data base that -- 3 Q. Let me ask you this: To your 4 knowledge, did anybody that participated in these 5 seventeen U.S. clinical trials that were analyzed 6 for the Beasley article attempt suicide longer 7 than a day after having been out of the trial? 8 A. I'm not certain. 9 Q. Di you ever make an attempt to 10 find that out? 11 A. I personally did not. 12 Q. Okay. Did you have an 13 understanding of what Lilly's policy was with 14 regards to following up or investigating events 15 surrounding a possible suicide attempt? 16 A. I'm not clear as to what their 17 actual policy is. 18 Q. What's your understanding of 19 what the policy was? 20 A. My understanding is that they 21 would apply all due diligence to follow up 22 patients, but I'm not sure exactly how that 23 policy is articulated. 24 Q. Doesn't that due diligence Page 256 1 include, at least theoretically, an investigation 2 of what other drugs that person might have been 3 on after discontinuing from a study? 4 A. I can't speak to that because 5 I'm not a physician, but whatever the physician 6 and the policy articulated, I'm sure that it 7 might include that, but I can't say exactly what 8 facets of that policy are there, whether it 9 includes other drugs or not. 10 Q. You don't have a problem with 11 them limiting the event to having to have 12 occurred one day after the end of their 13 participation in a clinical trial? 14 A. No. 15 Q. So this first person that's 16 listed in Exhibit Number 6, where it says study 17 drug was discontinued eight days prior to first 18 overdose, you don't have a problem with that 19 person being taken out of the analysis, at least 20 in this situation reflected in this exhibit, 21 because the study drug was discontinued eight 22 days prior to the first overdose? 23 MR. MYERS: Whatever the analysis is 24 now, that we've moved away from the Beasley Page 257 1 paper. 2 Q. Right, whatever the analysis 3 that this was done in preparation for, this 4 meaning Exhibit 6. 5 A. I personally do not have a 6 problem with this case. 7 Q. What was the definition of 8 discontinued from a trial? In other words, if a 9 person just never showed up after their fourth 10 visit again, would the fourth visit be considered 11 the date after which you determined whether or 12 not they're participating in the trial? 13 MR. MYERS: Let me object to the form. 14 For purposes of the Beasely paper or some other 15 analysis? 16 MS. ZETTLER: Any analysis at all. If 17 there's a difference, I want to know that, too. 18 A. I cannot speak to any 19 particular analysis. My impression is in 20 general, that a patient is lost to follow-up or 21 just never comes back, that a summary report is 22 written to close out that patient's case on that 23 date. 24 Q. Okay. Page 258 1 A. But I'm not sure because I'm 2 not familiar with that process. 3 Q. Okay. Let me ask it this way: 4 If a patient shows up for visit number four, 5 okay, as per the protocol and as per his 6 agreement with the clinical investigator, and 7 three days later they attempt suicide, but Lilly 8 doesn't find out about that attempt for another 9 month because nobody can get ahold of the 10 patient, is that patient taken into consideration 11 or considered a suicide attempt per the 12 definition or are they excluded because they 13 hadn't been, quote unquote, participating in the 14 trial for three days? 15 MR. MYERS: I object to the form, what 16 definition are you talking about? 17 MS. ZETTLER: The definition either 18 reflected in Exhibit 6 or the definition 19 reflected in Exhibit 5, the Beasley article. 20 A. Again, I'm not certain. 21 Q. Who would know? 22 A. The research physicians who are 23 the implementers of that process would know. 24 Q. Do you have any way of knowing Page 259 1 from the description of the first patient in 2 Exhibit Number 6 whether or not that person was 3 still on the study drug, still on Fluoxetine? 4 MR. MYERS: I object to the form, I 5 don't know that it discloses what drug they were 6 on. It doesn't -- you said -- you're assuming 7 Fluoxetine and all it says is study drug. 8 Q. Do you have any idea whether or 9 not that person was still on the study drug? 10 A. Please clarify it. On the 11 study drug at what point in time? 12 Q. When they tried to kill 13 themself? 14 A. It says here study drug was 15 discontinued eight days prior to first overdose. 16 So my reading of that is that they were not on 17 the study drug. 18 Q. And it also says under the 19 description of possible event, on 3-10-87 the 20 person tried to kill themself with twenty tablets 21 of an unknown anti-depressant, doesn't it? 22 A. Yes. 23 Q. You have no way of knowing 24 whether or not an attempt was made to determine Page 260 1 what that anti-depressant was, do you? 2 A. I don't know. 3 Q. Is it your understanding that a 4 clinical report form or clinical report forms 5 were filled out for this person who participated 6 in a study that is reflected as the first patient 7 in Exhibit 6? 8 A. To the best of my knowledge, 9 yes, because that's how we run clinical trials. 10 Q. Is there any way for you to 11 track, based on the information you have here, 12 this person back to the original written case 13 report form or computerized case report form? 14 A. If I had the full PIP or full 15 identification, I would be able to. 16 Q. In other words, would you agree 17 with me that a portion of the patient 18 identification has been marked out here? 19 A. Something has been marked out 20 under patient identification. 21 Q. Without that information, 22 assuming that that is part of the patient 23 identification, without that information you 24 would not be able to trace this back; correct? Page 261 1 A. Not necessarily. I could go 2 through every case and find a match ultimately. 3 Q. You could go through thousands 4 and thousands of cases and find a match? 5 A. I'm not sure how many cases, 6 but yes, ultimately I'm sure I could find a 7 match. 8 Q. But it would be much easier to 9 do it if you had the actual patient 10 identification number in complete form; correct? 11 A. Yes. 12 Q. To your knowledge was there a 13 concern at Eli Lilly regarding the incidence of 14 suicidality and the use of Fluoxetine prior to 15 1990 when Doctor Beasley's article was published -- 16 I'm sorry, Doctor Teicher's article was 17 published? 18 A. Not to my knowledge. 19 Q. To your knowledge was there 20 ever an analysis of the incidence of suicide 21 attempts and suicidal ideation done prior to the 22 analyses that were done, all of the analyses that 23 were done, in response to the Teicher article? 24 A. I'm not certain. Page 262 1 Q. Okay. Why do you say you're 2 not certain? 3 A. Because in order to get the 4 drug approved, for example, in the states, in the 5 United States, I know that safety analyses and 6 safety updates were provide to the Food and Drug 7 Administration, and among those analyses there 8 may have been one of suicide and suicidal 9 ideation, but I'm not sure whether those were 10 included or not. 11 Q. Have you been involved in a 12 safety analyses as far as adverse events have 13 been concerned in clinical trials for, say, I 14 believe you said you worked on obesity? 15 A. Yes. I have been involved with 16 analyzing safety from individual particular 17 clinical trials just as part of the final report 18 for that trial. 19 Q. Okay. Have you ever worked 20 with any outside consultants on the issue -- 21 except for suicide, we won't talk about suicide 22 right now -- on the issue of any particular 23 incidence of an adverse event? 24 A. No. Page 263 1 Q. Have you, in your experience at 2 Eli Lilly, known of there to be a policy at Lilly 3 when to hire outside consultants to help analyze 4 or make a determination with regards to the 5 incidence of any particular adverse events 6 besides suicidality? 7 A. I don't know of any consultants 8 who have been brought in just based on incidents 9 alone. We have consultants who have come in and 10 helped us with definition and just ascertaining, 11 you bring in medical experts to help us there. 12 Q. With regards to adverse events, 13 what types of definitions have you used outside 14 consultants for, say, the obesity or bulimia 15 trials? 16 A. I don't know. I don't know, 17 first of all, if we used consultants in those 18 particular trials or not, and secondly, if we 19 have, I don't know what adverse events they would 20 have been used for. 21 Q. I guess I'm a little confused 22 about what you meant with your prior answer that 23 you used consultants in some case other than 24 suicidality. Page 264 1 A. I'm sorry, I was referring to 2 in general, not obesity and bulimia. 3 Q. So other drugs? 4 A. Yes, just in general, the 5 policy is to bring in medical experts from times 6 to help out if we don't have the internal 7 expertise. 8 Q. Help with the definition? 9 A. Definition or just help sort 10 out what this is saying, what the meaning of 11 things is. 12 Q. As far as determining or 13 helping to determine the incidence of an adverse 14 event, though, other than suicidality, it's not 15 been your experience that Lilly would use outside 16 consultants? 17 MR. MYERS: With what? 18 MS. ZETTLER: With regards to adverse 19 events. 20 A. No, that's not correct. As I 21 said, in general we have, and I can't think of 22 specifics, but brought in consultants to help 23 with various adverse events. 24 Q. Other than writing definitions, Page 265 1 have you used them to -- and I'm talking outside 2 of suicidality. 3 A. I'm not sure because I haven't 4 been primarily involved in those discussions 5 because they're between physician and physician 6 or pharmacologist. 7 (PLAINTIFFS' EXHIBIT NO. 7 WAS 8 MARKED FOR IDENTIFICATION AND 9 RECEIVED IN EVIDENCE.) 10 Q. Have you had a chance to review 11 Exhibit Number 7? 12 A. Yes. 13 Q. Do you recognize that exhibit? 14 A. Yes. 15 Q. Can you tell me what this 16 exhibit reflects? 17 A. It reflects a summary of the 18 suicide attempt rates and the time course of 19 attempts for various patient populations that 20 were agreed upon in a London meeting of March 21 21st, 1985. 22 Q. Were you involved with this 23 project at all? 24 A. No. Page 266 1 Q. Why is it that it was something 2 that was in your file? 3 A. I took over Fluoxetine 4 responsibilities for obesity and bulimia, and 5 David Hardison was a statistician in our group 6 who is no longer with the company, and those 7 files were transferred to another statistician, I 8 got some of them and possibly Bruce Dornseif got 9 some. 10 Q. You have read this before? 11 A. Yes, I believe so. 12 Q. Do you remember why this 13 summary or this analysis was being done back in 14 1985? 15 A. I was not involved with 16 Fluoxetine at that time, so I don't know the 17 details behind this. 18 Q. Okay. Do you know if it was 19 being done for a submission to the FDA? 20 A. I can't determine that. 21 Q. Have you ever seen an FDA 22 document in which the results of this summary 23 were reported? 24 MR. MYERS: When you say an FDA Page 267 1 document, what document? 2 MS. ZETTLER: Documents submitted to 3 the FDA by Eli Lilly. 4 A. Not to my knowledge. 5 Q. Are you familiar with why there 6 was a meeting in London held on March 21, 1985? 7 A. No, I'm not. 8 (PLAINTIFFS' EXHIBIT NO. 8 WAS 9 MARKED FOR IDENTIFICATION AND 10 RECEIVED IN EVIDENCE.) 11 Q. Is Exhibit Number 8 related to 12 Exhibit Number 7 in any way? 13 A. I'm not sure. 14 Q. Are you familiar with Exhibit 15 Number 8? 16 A. I don't believe so. 17 Q. Why is this -- why, to your 18 knowledge, would this have been produced as a 19 document from your files? 20 A. My files included everything 21 that David, for example, gave to me when he left 22 the company, so it may have been part of that 23 file. 24 Q. Okay. Have you ever seen a Page 268 1 printout similar to the printout reflected in 2 Exhibit 8? 3 A. I've seen printouts that look 4 similar. 5 Q. Okay. Can you tell is there a 6 common name for this type of a printout within 7 Lilly or is this something that was -- I'm trying 8 to figure out what we can refer to it as instead 9 of saying the computer printout. 10 A. This looks to me like a patient 11 listing. 12 Q. Okay. And according to the 13 exhibit, it's a listing of patients who suffered 14 suicide attempts; correct? 15 MR. MYERS: Are you reading off the 16 document somewhere? 17 MS. ZETTLER: At the top of the 18 document it says Fluoxetine clinical trials dash 19 suicide attempt. What I'm trying to find out is 20 this patient -- these patients represented 21 patients who suffered suicide attempts. 22 A. I can only take verbatim what 23 the title says here, I don't know if this Exhibit 24 here is one patient or multiple patients, too, I Page 269 1 can't determine that. 2 Q. I think it's two patients in 3 this particular packet. If you look at the age 4 and height on the left-hand side, the first 5 patient was fifty-six for age and height, 6 sixty-four, and then I believe towards the back 7 of the exhibit, those ages change from 8 fifty-three to sixty-four. And again, you can't 9 really tell for sure if these are the same 10 patients or different patients because the 11 investigator number and the patient number are 12 blacked out; correct? 13 A. It's not easy to determine 14 that. 15 Q. So what we have to assume that 16 these are two different patients because of the 17 difference in the age; correct? 18 A. Possibly. 19 Q. As it's listed. It also could 20 reflect a change or a correction made to the age; 21 correct? 22 A. I don't know. 23 Q. Were similar patient listings 24 run on suicide attempts in preparation for the Page 270 1 analysis of the Beasley article or for analysis 2 that was submitted to the advisory committee? 3 A. I'm not certain. 4 Q. Who would know that? 5 A. The systems personnel generally 6 run patient listings. 7 Q. Who did you work with on the 8 Beasley article from systems? 9 A. I personally did not work with 10 any particular individual in systems because I 11 wasn't involved in the day-to-day analysis, but 12 one of the authors on the BMJ paper is a systems 13 analyst. 14 Q. Who was that? 15 A. David Murphy. 16 Q. So that would indicate that he 17 did significant work in preparation for that 18 article? 19 A. It would indicate he did work 20 in preparation for that article, I can't 21 determine the significance -- you know, what 22 significance -- I'm not sure how much work he did 23 for the article. 24 Q. Is it Lilly's practice to name Page 271 1 people as authors of articles who did minimal 2 work on a project? 3 A. It's Lilly's -- no, it's 4 Lilly's practice to name people who did work that 5 contributed significantly to the science and the 6 process of actually getting the article prepared. 7 MR. MYERS: Off the record a second. 8 (DISCUSSION OFF THE RECORD.) 9 (PLAINTIFFS' EXHIBIT NO. 9 WAS 10 MARKED FOR IDENTIFICATION AND 11 RECEIVED IN EVIDENCE.) 12 Q. Have you had a chance to review 13 Exhibit 9? 14 A. Yes. 15 Q. Is this your handwriting? 16 A. No. 17 Q. Do you know whose handwriting 18 this is? 19 A. Yes. 20 Q. Whose? 21 A. Bruce Dornseif. 22 Q. Let me ask you this before I 23 forget: You said David Hardison is no longer 24 with Lilly? Page 272 1 A. Yes. 2 Q. Do you know when he left Lilly? 3 A. No, I do not. 4 Q. Do you know where he is today? 5 A. I'm not certain. 6 Q. Okay. Where do you think he 7 may be? 8 A. The last I heard is that he was 9 in North Carolina. 10 Q. Do you know what he was doing 11 in North Carolina? 12 A. I'm not certain. 13 Q. Okay. What do you think he was 14 doing in North Carolina? 15 A. Well, he is a statistician, so 16 I'm assuming he's doing something related to what 17 statisticians do. 18 Q. Where was he working that you 19 last heard of? 20 A. Okay. I knew that leaving 21 Lilly he went, I believe, to Vanderbilt to work 22 in the business department there or some -- some 23 department at Vanderbilt. 24 Q. To your knowledge, has he left Page 273 1 Vanderbilt? 2 A. Yes. 3 Q. When did he leave Vanderbilt? 4 A. I do not know. 5 Q. Do you know where he went after 6 Vanderbilt? 7 A. I believe, I'm not certain here 8 because I haven't followed up on his career, but 9 I believe he went to either Hospital Corporation 10 of America or some large health HMO, but I'm not 11 sure. 12 Q. Have you seen Exhibit Number 9 13 before? 14 A. I'm not certain. 15 Q. Did you participate in this 16 meeting, 6-26-87, suicide Fluoxetine meeting? 17 A. Most likely no. 18 Q. Have you ever participated in a 19 Fluoxetine suicide meeting? 20 MR. MYERS: What do you mean by 21 Fluoxetine suicide meeting? There's a lot of 22 discussion about this BMJ and reviewing it. 23 Q. Where actually individual cases 24 were reviewed. Page 274 1 A. Not to my knowledge. 2 Q. You mean not to your memory? 3 A. Yes, I don't remember reviewing 4 individual cases. 5 Q. If you had been there, you 6 would have knowledge of it at the time, but you 7 have forgotten; correct? 8 A. Yes, I don't remember. 9 Q. Okay. Have you ever attended a 10 meeting where individual cases of any adverse 11 events were reviewed? 12 A. I have been in meetings where 13 adverse events in patients have been discussed. 14 Q. In particular patients or in 15 general? 16 A. In general. 17 Q. You've never sat down with 18 somebody from the medical department and reviewed 19 an individual case report to decide whether or 20 not a person had suffered from an adverse event? 21 A. No. 22 (PLAINTIFFS' EXHIBIT NO. 10 WAS 23 MARKED FOR IDENTIFICATION AND 24 RECEIVED IN EVIDENCE.) Page 275 1 Q. Have you had a chance to look 2 at Exhibit 10, Doctor? 3 A. Yes. 4 Q. Do you recognize this exhibit? 5 A. I believe I have seen it 6 before. 7 Q. That you have seen it before? 8 A. Yes. 9 Q. Is this related to other 10 exhibits that we have just looked at? 11 MR. MYERS: Starting with which one? 12 Q. Such as Exhibit Number 7? 13 A. I'm not certain, but I believe 14 so, it's in the same time frame, same issue. 15 Q. On the first page of that 16 Exhibit Number 10, second paragraph, the writer 17 states these data were relevant to the question 18 of whether or not Fluoxetine was associated with 19 an increased number of suicide attempts when 20 compared to placebo controls and comparitor drug 21 controls. Again this was back in 1985? 22 A. Yes. 23 Q. Does this refresh your 24 recollection as to whether or not Lilly was Page 276 1 concerned about an increased incidence of 2 suicidality and Fluoxetine? 3 MR. MYERS: At that point in time? 4 MS. ZETTLER: Yes. 5 A. It shows to me that they were 6 looking into it. 7 Q. Do you know if this was because 8 somebody at Lilly noticed that there was a higher 9 incidence, at least upon review of clinical 10 report forms? 11 A. I'm not certain. 12 Q. Has anybody ever told you that 13 suicidality and the use of Fluoxetine had been an 14 issue of concern at Lilly for a long time prior 15 to 1990? 16 A. No. 17 Q. Has anybody ever told you that 18 it has not been an issue of concern at Eli Lilly 19 prior to 1990? 20 A. No. 21 Q. On the third page of Exhibit 10 -- 22 actually, you have the luxury of having page 23 numbers in the upper left-hand corner this time 24 so it should make it a little easier. Under Page 277 1 number five it says as the half-life of the drug 2 is three to four days, would you agree that this 3 in fact could be related to the drug; correct? 4 A. Yes. 5 Q. Earlier you testified that it 6 was your belief that the half-life was only one 7 day? 8 A. Yes. 9 Q. Or two days? 10 A. Yes. 11 Q. Does this change your belief? 12 A. It may. I don't know if 13 they're correct or not. Again, like I said, I'm 14 uncertain. 15 Q. Do you know who these people 16 were, people that wrote this article, the person 17 that wrote this report? 18 A. They're all blacked out. 19 Q. You don't have any recollection 20 as to who these people may have been? 21 A. No. 22 Q. Without looking at -- without 23 having the names or having these names blacked 24 out you can't tell who wrote this; correct? Page 278 1 A. I can tell a bit. There's one 2 Ph.D and two M.D.'s still listed there, so. 3 Q. Have you ever worked with a 4 group of two M.D.'s and one Ph.D at Lilly as 5 consultants -- not at Lilly, but consultants, 6 have you ever worked with a group such as that? 7 MR. MYERS: With respect to this? 8 MS. ZETTLER: With respect to anything. 9 A. Have I worked with M.D.'s and 10 Ph.D's or exactly two? 11 Q. Exactly two. A group of three 12 people comprising two M.D.'s and one Ph.D. 13 A. I have not worked with any such 14 group directly. 15 Q. And on page seven of the 16 report, the authors talk about, at the bottom of 17 the first paragraph, in any event the 18 significance favoring the fact that Fluoxetine 19 was associated with more attempts varied with P 20 numbers of, looks like, point oh five one, point 21 oh five six and point one nine. Has that been 22 your experience, that the numbers or the 23 significance favoring the fact that Fluoxetine 24 was associated with more attempts varied with P Page 279 1 numbers? 2 A. I can only state what's given 3 here, there's three P values given, and they 4 varied between point one nine and what appears to 5 be point five one, I can't be for sure. 6 Q. It's hard to read with that big 7 stamp across the middle of it, isn't it? 8 A. Yes. 9 Q. When you did your analysis or 10 people within your group did their analysis of 11 the data for the Beasley article, did you do that 12 analysis with varying P numbers or did you stick 13 with just the .05 number? 14 A. There's two different issues 15 here. 16 Q. Okay. 17 A. .05 is the level of 18 significance, set a priorty, you don't calculate 19 that, that's a given level. 20 Q. What's the difference between -- 21 that's a P number, isn't it? 22 A. It's a significance level. 23 Then you calculate P values and determine whether 24 those are larger or smaller than .05. So .05 is Page 280 1 set up front as the decision cutoff, and then you 2 go ahead and calculate P values. If P values are 3 less than .05, you state it's statistically 4 significant, if they're greater than .05, you 5 don't state it's statistically significant. 6 Q. What's the difference between 7 the P numbers that I've talked about in Exhibit 8 10 and the P number of .05? 9 A. The P number of .05 is really 10 not a calculated P number, it is a criteria, it's 11 a threshold value that's set, it's not 12 calculated, it's a value that's set. And then 13 you calculate these P values or P numbers, as you 14 say, and you determine whether they're less than 15 .05, this predefined number, or greater than .05. 16 All of these three P values here are greater than 17 .05. .05 is set beforehand, and these three 18 numbers are greater than .05. 19 Q. Is there any way to tell 20 whether or not they set the P number before .05? 21 MR. MYERS: In this exercise? 22 MS. ZETTLER: Right, in this exercise. 23 A. I don't see one way or the 24 other whether they -- what significance level Page 281 1 they were using. 2 Q. Would you have -- would they 3 have to set a significance level to calculate 4 these P numbers reflected in Exhibit 10? 5 A. No, you do not have to set a 6 significance level to calculate these P numbers 7 because -- 8 Q. Okay. Do you have a 9 recollection of what P numbers were calculated 10 during the analysis of the Beasley article data? 11 A. I can refer to the article 12 here. 13 Q. Sure, go ahead. 14 A. There are P values listed in 15 the abstract on the first page, for example. 16 Q. Okay. So you're talking about 17 in, I believe, in starting in the results section 18 at the bottom of the left-hand side? 19 A. Yes. 20 Q. It says suicidal acts did not 21 differ significantly in comparisons of Fluoxetine 22 with placebo, paren, zero point two percent 23 versus zero point two percent, comma, P equals 24 zero point four nine four; correct? Page 282 1 A. Right. 2 Q. Can you tell from Exhibit 10 3 what statistical analysis they used on the data? 4 A. I'm not sure exactly how, for 5 example, these three P values were calculated. 6 Two methods are suggested in that paragraph, one 7 is Fischer's exact probability, and then doctor 8 somebody suggested using a binomial inspection, 9 and I'm not sure exactly what method led to these 10 three particular P values. 11 Q. But you feel certain that the P 12 numbers that they're talking about in Exhibit 10 13 are not the significance of the P numbers, the 14 .05 numbers? 15 A. These P numbers are P values 16 that they calculated somehow based on the data. 17 The level of significance is not calculated based 18 on any particular study data, it's a criteria set 19 up independent of the data analysis. 20 Q. But you can set up a criteria 21 other than .05, can't you? 22 A. Right. 23 Q. You can use point one; correct? 24 A. Yes. Page 283 1 Q. You can use point oh five one 2 if you wanted to, couldn't you? 3 A. If someone wanted to, but most 4 reasonable scientists who analyze data this way 5 had traditionally accepted that P -- the 6 significance level of .05. 7 Q. Is this something that you 8 learned while at Lilly or something you learned 9 throughout your education? 10 A. Throughout my education. 11 Q. I guess -- because I'm a little 12 confused here because I don't -- it says here 13 significance favoring the fact that Fluoxetine 14 was associated with more attempts varied with P 15 numbers. If those are numbers that are 16 calculated as a result -- that's just what that 17 suggests to me, okay, and granted I'm a regular 18 lay person at this stuff, but it seems to me that 19 they're using these P numbers to calculate, these 20 aren't P numbers that they're getting as a result 21 of the calculation? 22 A. Yes, they are. I'm not sure 23 what that calculation is, I'm suggesting there's 24 two methods given here. Some method was used to Page 284 1 calculate these P numbers. 2 Q. But these are numbers that 3 they're getting as a result of a calculation? 4 A. Yes, and these are called 5 significance values, but they aren't significant 6 if they're using the level of .05. But they are -- 7 people refer to these P values also as 8 significance numbers, as they do here. 9 Q. But the way the sentence reads 10 is that in any event the significance favoring 11 the fact that Fluoxetine was associated with more 12 attempts varied with P numbers. So they're 13 saying that the P numbers are the significance? 14 A. The P numbers are the values 15 that you have to compare to some level of 16 significance, which traditionally has been .05. 17 But these numbers are varying. So the 18 significance as to how close they are to that 19 threshold level is varying, some are close and 20 some are farther away. 21 Q. They're saying that the 22 variance is between the point oh five one and 23 point one nine? 24 A. Yes. Page 285 1 Q. The those three numbers reflect 2 variances. 3 A. Yes. Now I'm not clear as to 4 how each particular number came about, whether 5 they used different methods or data sets, I'm not 6 sure right now, so I'm not sure why they're 7 varying, but they throw three numbers out here 8 and they are different from each other. 9 Q. You're convinced that these are 10 not the P values that we talked about earlier, 11 these are not numbers that they used to 12 calculate? 13 MR. MYERS: They used or came up with? 14 MS. ZETTLER: Used. 15 Q. In other words, like the P 16 number of .05, these are not those types of 17 numbers, those are results? 18 A. These are results listed right 19 here. This .05 is not a result, it's an A 20 priority stated criteria. 21 MS. ZETTLER: Let's take a lunch break. 22 (A LUNCH BREAK WAS TAKEN.) 23 (PLAINTIFFS' EXHIBIT NO. 11 WAS 24 MARKED FOR IDENTIFICATION AND Page 286 1 RECEIVED IN EVIDENCE.) 2 Q. Have you had a chance to look 3 at Exhibit Number 11? 4 A. Yes. 5 Q. Do you recognize this exhibit? 6 A. Yes. 7 Q. Can you tell me what it is? 8 A. It's a protocol. 9 Q. For what? 10 A. In depth assessment of patients 11 with psychobehavioral deterioration while on 12 anti-depressants. 13 Q. Is this a rechallenge? 14 A. Not as I understand 15 rechallenge. 16 Q. What's the purpose of doing 17 this study, if you know? 18 A. Just what it says here. In the 19 overall design and plan, the primary objective of 20 this study is to more clearly characterize the 21 variety of phenomena constituting 22 phychobehavioral deterioration observed in 23 patients being treated with antidepressants. 24 Q. And that's -- are you reading Page 287 1 from -- 2 A. Page one. 3 Q. Letter A, overall design and 4 plan? 5 A. Yes. 6 Q. Did you work on this protocol? 7 A. I'm not sure. 8 Q. Would you look on page four of 9 Exhibit 11, please, and at the bottom it says, 10 under category L, statistical and analytical 11 plans, each investigator will develop a report to 12 characterize -- and characterize the range of -- 13 or categorize, I'm sorry, the range of phenomena 14 as exhibited by the patients referred to him. 15 These reports will be submitted to a data 16 monitoring board assigned by Eli Lilly and 17 Company. A final report of the overall 18 characterization of phenomena of psychobehavioral 19 deterioration while on antidepressants will be 20 based upon the recommendation of the board's 21 assessment of the reports; correct? 22 A. Correct. 23 Q. That doesn't say anything about 24 a statistical analysis, does it, a specific Page 288 1 statistical analysis? 2 A. It does not refer to a specific 3 statistical analysis. 4 Q. To your knowledge, was a 5 specific statistical analysis decided upon before 6 this protocol was written? 7 A. I'm not sure. 8 Q. Is it your experience at Eli 9 Lilly that statistical analyses are determined or 10 developed prior to the running of a clinical 11 trial? 12 A. In general, yes. But this -- 13 as I read this protocol here, it's not a typical 14 clinical trial, it's a characterization of a 15 patient population, there's no treatments being 16 compared, et cetera. For any protocol where we 17 do prepare treatments, we always have an A 18 priority prespecified analytical plan. This is 19 not that type of clinical trial, so that's why 20 the general characterization is mentioned here, 21 overall characterization. 22 Q. Was a protocol written for the 23 meta analysis done by Doctor Beasley for his 24 article in the BMJ? Page 289 1 A. I'm not sure if one was written 2 for that, that was a retrospective study and 3 prospective studies always have a protocol that 4 people follow, and retrospective analyses are not 5 referred to using, quote, a traditional protocol. 6 Q. Do you know if something like 7 protocol was written prior to the retrospective 8 analysis? 9 A. Various methods and analyses 10 were stated and documented. Things were written 11 down regarding analysis, for example, just the 12 algorithm that the physicians used to classify 13 patients. So various components that would be 14 included in the protocol were written down 15 beforehand. Whether you would call that all a 16 protocol, I'm not sure, I don't know. 17 Q. Is this study that's reflected 18 in Exhibit 11, was this a study that was to be 19 performed in preparation for the rechallenge 20 protocol, rechallenge study? 21 A. I'm not sure, I'm not sure if 22 it was in preparation for or to be run 23 simultaneously, I'm not sure. 24 Q. The study that we talked about Page 290 1 yesterday, as reflected in Exhibit 2, was a study 2 to validate a scale that was to be used in the 3 rechallenge study; correct? 4 A. Correct. 5 Q. Can you tell from looking at 6 Exhibit 2 and comparing it to Exhibit 11 whether 7 or not those two studies were to be -- both to be 8 run for purposes of use in the rechallenge study 9 or in preparation for the rechallenge study? 10 A. The Exhibit 2 was done in 11 preparation for a rechallenge study because it 12 was validating the instrument that was to be 13 used. I'm not sure about Exhibit 11, whether 14 again that was to be done prior to or 15 simultaneously or whatever. 16 Q. But you said a couple of 17 minutes ago that the purpose of this study that's 18 reflected in Exhibit 11 was to characterize a 19 patient population; correct? 20 A. Correct. 21 Q. Okay. And that patient 22 population, at least from the title, is a 23 population of patients who had psychobehavioral 24 deterioration while on antidepressants; correct? Page 291 1 A. I'm not sure if it refers to 2 patients who had psychobehavioral deterioration 3 or currently were deteriorating. 4 Q. Okay. 5 A. On antidepressants. 6 Q. But regardless of when they 7 were experiencing the deterioration, it was 8 involving patients who experienced 9 psychobehavioral deterioration while on 10 antidepressants? 11 A. Correct. 12 Q. What you know about the 13 proposed rechallenge protocol or from what you 14 know about the proposed rechallenge protocol, 15 would this be a patient population that may be 16 included in that rechallenge protocol? 17 A. I'm not sure. 18 Q. Does this Exhibit 11 say 19 anything specifically about suicidal ideation or 20 suicidality? 21 A. My reading of this does not 22 show any specific term suicidality or suicidal 23 ideation, however I do know that some of the 24 scales being used do measure those phenomena. Page 292 1 Q. And you're talking about the 2 scales that are listed on page three under 3 efficacy data, and the subsections of A, B? 4 A. Correct. 5 Q. And which scales are you 6 talking about? 7 A. For example MSSIR. 8 Q. Okay. Any others? 9 A. Item three, the Ham-D 10 twenty-eight includes, I believe, the item three 11 which we have discussed. 12 Q. Any others? 13 A. I can't speak to any of the 14 others. 15 Q. Do you know if the hopelessness 16 scale is related to suicidality? 17 A. I'm not sure in which way, how 18 it's related to suicidality. So no, I'm not 19 certain. 20 Q. Do you remember in what context 21 you read this exhibit prior to today, in other 22 words in relationship to what projects, if any? 23 A. I'm not certain in what context 24 I read it because I'm not certain whether I read Page 293 1 this particular thing. I have a recollection of 2 either seeing or hearing something about a study 3 which characterizes these types of patients. 4 Whether I read this actual draft at this stage, 5 I'm not certain. 6 Q. Do you know if this study was 7 ever performed? 8 A. No, I do not. 9 Q. Do you know that it was not 10 performed? 11 A. No, I do not. 12 (PLAINTIFFS' EXHIBIT NO. 12 WAS 13 MARKED FOR IDENTIFICATION AND 14 RECEIVED IN EVIDENCE.) 15 Q. Okay. Could you look at 16 Exhibit, please. 17 A. (THE WITNESS COMPLIES.) 18 Q. Have you had a chance to review 19 Exhibit 12? 20 A. Yes. 21 Q. Can you tell us what that is? 22 A. That is a protocol. 23 Q. Do you recognize it? 24 A. Yes. Page 294 1 Q. Can you tell us what it's a 2 protocol for? 3 A. It's a protocol outlining the 4 comparitor study between Fluoxetine, Desipramine 5 and placebo in patients who developed suicidal 6 ideation while under treatment. 7 Q. Is this the rechallenge 8 protocol? 9 A. Yes. 10 Q. And did you do any work in 11 developing this protocol? 12 A. Yes. 13 Q. Tell me what you did? 14 A. I calculated sample size under 15 sample size criteria on page four. 16 Q. Okay. Anything else? 17 A. And I helped write the 18 statistical and analytical plans, section L, page 19 fourteen and fifteen. 20 Q. If we can go back to page four. 21 It says at the bottom of that section three, if 22 less than thirty patients are randomized in this 23 study in the first year, the study will be 24 stopped and the concept re-evaluated; correct? Page 295 1 A. Correct. 2 Q. Do you know if that happened, 3 if they had to re-evaluate the concept? 4 A. No, I do not know. 5 Q. Did anybody else work on this 6 calculation of sample size that you did? 7 A. I'm not sure. I know I, for 8 one, made a calculation, I'm not sure if others 9 made a similar calculation or not. 10 Q. Is this your calculation that 11 they used in this draft? 12 A. To the best of my memory, yes, 13 at least I came up with those numbers. They may 14 have used someone else's with the same number, 15 but I believe this looks like my calculation. 16 Q. So you wanted eighty patients, 17 you needed eighty patients ramdomized, total; 18 correct? 19 A. Correct. 20 Q. And per your calculations, if 21 you randomized eighty patients, sixty patients 22 would complete the initial four-week treatment 23 period? 24 A. That is what was expected, that Page 296 1 if you randomized eighty, sixty would complete. 2 Q. Based on what? 3 A. That's a medical determination, 4 just looking at the natural course of the disease 5 and previous patient experience. 6 Q. So was that number -- was that 7 a number that you came up with, the ramdomizing 8 eighty, or is that a number that they determined 9 in the medical department? 10 A. That is not a statistical 11 number, that's not my number, it's based on prior 12 medical experience, how many you would expect to 13 complete four weeks of treatment of these 14 particular patients. 15 Q. Okay. If you could turn to 16 page fifteen, under number three, it says data 17 monitoring in interim analysis. The study will 18 be monitored by a data monitoring board. Do you 19 know if any such board was ever established? 20 A. No, I do not. 21 Q. Do you know if anybody was 22 being considered for participation on that board? 23 A. No, I do not. 24 (PLAINTIFFS' EXHIBIT NO. 13 WAS Page 297 1 MARKED FOR IDENTIFICATION AND 2 RECEIVED IN EVIDENCE.) 3 Q. Could you look at Exhibit 13, 4 please? 5 A. (THE WITNESS COMPLIES.) 6 Q. This looks an awful lot like 7 Exhibit Number 11, would you agree? 8 A. Yes. 9 Q. Is it the same protocol? 10 A. It looks an awful lot like 11 Exhibit 11. The protocol PIP or project code is 12 changed, but I don't know if it's the same -- 13 it's definitely not the same, it's not identical. 14 Q. So it could be two separate 15 studies? 16 A. It could be, but my guess is 17 given the same title and same objectives and 18 everything, it looks rather the same, that's the 19 same study. 20 Q. Do you have any explanation why 21 the study numbers would be different? 22 A. No, I do not. 23 Q. Do you see any differences 24 between the two besides the study number? Page 298 1 A. Well, there are -- they are 2 different. 3 Q. In what way? 4 MR. MYERS: Nancy, before he answers, 5 you know, I don't know that it's fair to make him 6 go through and compare them line by line. If he 7 can point out some differences, that's fine, but 8 I don't want him to be bound unless we're willing 9 to go through it line by line as to what the 10 differences are. Obviously there -- 11 Q. I'm not talking word for word, 12 I'm talking about substantively, if you know of 13 any substantive differences. 14 A. No, I do not. 15 Q. Like for instance sample sizes, 16 statistical analyses. I'm just trying to make 17 sure as best we can that this is not two separate 18 studies. 19 A. The sample size wording is 20 identical in both. 21 Q. Okay. 22 A. The statistical and analytical 23 plan is identical in both. 24 Q. Could you look at the third Page 299 1 page of Exhibit 13? 2 A. Okay. 3 Q. Is that your handwriting on 4 that page? 5 A. I don't believe so. 6 Q. Do you recognize the 7 handwriting? 8 A. No, I don't. 9 Q. It looks like there's almost 10 two different handwritings on that page, doesn't 11 it, unless one's just bigger than the other? 12 A. I can't determine that. 13 Q. How about on the next page, is 14 that your handwriting on the next page? 15 A. I'm not sure, it looks more 16 like my handwriting than the other handwriting, 17 but I can't be certain. 18 Q. Have you ever spoken with Dr. 19 Teicher, Dr. Martin Teicher? 20 A. No, I have not. 21 Q. Has anybody from your section 22 attempted to work with Dr. Teicher to come up 23 with the statistical analysis for the rechallenge 24 protocol or any of the protocols done in Page 300 1 connection with the rechallenge protocol? 2 A. You mean my own statistics 3 group? 4 Q. Right. 5 A. I don't believe that anyone 6 within our group has attempted to speak with Dr. 7 Teicher. 8 Q. Has anybody from Eli Lilly, to 9 your knowledge, attempted to speak with Dr. 10 Teicher? 11 A. I'm not sure. 12 (PLAINTIFFS' EXHIBIT NO. 14 WAS 13 MARKED FOR IDENTIFICATION AND 14 RECEIVED IN EVIDENCE.) 15 Q. Okay. First of all, yesterday 16 we talked about whether or not you had attended 17 meetings at the FDA or with the FDA regarding 18 rechallenging and you said you recall one meeting 19 and, I believe, that was reflected in Exhibit 1, 20 the May, 1991 meeting? 21 A. Right. 22 Q. Does this exhibit refresh your 23 recollection as to whether or not you attended 24 any other meetings with the FDA regarding Page 301 1 rechallenge? 2 A. Yes. 3 Q. Do you have a specific -- 4 A. Let me correct that. 5 Q. Okay. 6 A. This meeting was to discuss 7 postmarketing safety, and rechallenge was a part 8 of that meeting. 9 Q. It wasn't specifically a 10 meeting specifically convened to discuss the 11 issue of rechallenge? 12 A. That's correct. 13 Q. Does this Exhibit Number 14 14 refresh your recollection as to whether or not 15 the FDA requested that Lilly perform a 16 rechallenge study on the issue of suicidality? 17 A. It does not state that the FDA 18 requested a rechallenge study. 19 Q. But does it refresh your 20 recollection as to whether or not they actually 21 requested that a rechallenge study be performed? 22 A. No, all I remember is what it 23 says here, that that was discussed. 24 Q. Is this the first time that you Page 302 1 heard of rechallenge with regards to the issue of 2 suicidality and the use of Fluoxetine? 3 A. I'm not sure. 4 Q. Do you remember being at this 5 meeting? 6 A. Yes. 7 Q. Do you remember -- do you know 8 if this exhibit should be more than one page 9 long? 10 A. I'm not sure. 11 Q. It kind of looks like there's a 12 staple mark in the upper left-hand corner, 13 doesn't it? 14 A. There is a line in the upper 15 left-hand corner, I'm not sure what that is. 16 Q. Have you ever seen this Exhibit 17 before? 18 A. I'm not sure. 19 Q. Is this an -- are these meeting 20 minutes minutes that were created by somebody at 21 Lilly or are these meeting minutes created by 22 somebody at the FDA to your knowledge? 23 A. I can't be certain. 24 Q. Okay. Page 303 1 A. It doesn't look like something 2 that was generated at Lilly just given the font 3 and the way it's laid out, but I cannot be 4 certain. 5 Q. Do you recall talks with Dr. 6 Teicher at Lilly or through Lilly regarding the 7 operational criteria for his phenomena? 8 A. No, because our statistical 9 group wouldn't be involved in any type of 10 deliberation like that. 11 Q. To your knowledge, did Dr. 12 Teicher do a statistical analysis in the paper 13 that he published or for the paper that he 14 published? 15 A. I'm not certain if he did a 16 statistical analysis. 17 Q. Do you have any knowledge as to 18 the validity of different types of clinical 19 trials? In other words -- I should say -- let me 20 rephrase that. Do you have any knowledge about 21 the reliability of different types of rechallenge 22 or different types of clinical trials -- it's 23 been a long week. 24 A. I don't understand what you're Page 304 1 asking. 2 Q. In other words, as a 3 statistician, would you have more confidence in 4 the data that was gained through a rechallenge or 5 a perspective type study or a retrospective type 6 study? 7 A. I would have most confidence in 8 a randomized controlled prospective study. 9 Q. Okay. Whether or not it was a 10 rechallenge type study? 11 A. That's correct, a randomized 12 prospective controlled study. 13 Q. Did you -- with regards to the 14 issue of suicidality, did you have an objection 15 to doing a retrospective analysis of data that 16 had been collected through the clinical trials as 17 opposed to setting up a prospective study to 18 study the issue of suicidality? 19 A. No. 20 Q. Why not? 21 A. Because the retrospective 22 analysis was done on prospective randomized 23 controlled clinical trials. 24 Q. Okay. Page 305 1 (PLAINTIFFS' EXHIBIT NO. 15 WAS 2 MARKED FOR IDENTIFICATION AND 3 RECEIVED IN EVIDENCE.) 4 Q. Have you had a chance to look 5 at Exhibit 15? 6 A. Yes. 7 Q. Do you recognize this exhibit? 8 A. I don't recognize it, I was an 9 addressee. I have read it, but I don't remember 10 details from this particular exhibit. 11 Q. Do you recall the subject 12 matter of the exhibit? 13 A. Yes. 14 Q. Can you tell me what you recall 15 about the subject matter? 16 A. The authors were laying out 17 three approaches to the study of potential 18 association of suicidality to antidepressant 19 therapy. The first would be an epidemiological 20 approach, the second would be a prospective 21 randomized control clinical trial, and the third 22 a ramdomized rechallenge study. 23 Q. And they listed them from least 24 favorable or least reliable to most reliable in Page 306 1 their minds, didn't they? 2 MR. MYERS: Before he answers, let me 3 object to what or wasn't in the minds of the 4 authors of this, and if you're asking him if he 5 could tell the order in which they were listed, I 6 think that's okay. 7 Q. Can you tell which order they 8 were listed in? 9 A. Yes, I can. 10 Q. What order were they listed in? 11 A. Epidemiologic studies first, 12 large prospective double-blind parallel 13 comparison trials, second, and a rechallenge 14 study, third. 15 Q. And according to this exhibit, 16 they felt that the rechallenge study would be the 17 most -- would be the best option of the three; 18 correct? 19 A. They said is considered the 20 best of the three options. 21 Q. And the epidemiology study 22 would be the least favorable approach? 23 A. They do not state that here. 24 Q. Why don't you do this: Take a Page 307 1 couple of minutes and read this entire first 2 three pages of this and then we'll talk about it. 3 A. Okay. 4 Q. Would you agree that they list 5 the epidemiology study first? 6 A. Yes. 7 Q. And they talk about various 8 disadvantages to running an epidemiology study; 9 correct? 10 A. Yes. 11 Q. And an epidemiology study, at 12 least reflected in this exhibit, is a 13 retrospective analysis of data that had been 14 collected previously; correct? 15 A. They mentioned a retrospective 16 cohart study design or retrospective case control 17 study design. I'm not sure that it's entirely 18 just retrospective and I'm not certain where 19 there might be a prospective component to that, 20 too. 21 Q. But they don't mention a 22 prospective component to an epidemiology study, 23 do they? 24 A. Not explicitly. Page 308 1 Q. Did they do it implicitly, 2 where do they imply that? 3 A. Not here, it's just that in my 4 little knowledge of epidemiology, sometimes these 5 case control studies can include a prospective 6 component as well, but they didn't explicitly 7 mention that in regard to that here. 8 Q. In the second -- 9 A. May I -- let me go ahead and 10 finish there. I'm referring to the last sentence 11 of the second paragraph. 12 Q. Right. 13 A. Using a historical prospective 14 study design or retrospective cohort. So at 15 least implicitly there there's something with 16 respect to retrospective plus prospective, 17 together. 18 Q. But it also goes on to say -- 19 the entire sentence says: Using a historical 20 prospective retrospective cohort study design or 21 a retrospective case control study design would 22 require previously gathered data; correct? 23 A. Yes, both would. But the first 24 one includes this prospective cohort as well. Page 309 1 See, these are two types of epidemiologic 2 studies. I'm not an epidemiologist, but they are 3 two separate kinds of studies. 4 Q. But what you're saying is it 5 would be, in the case of the historical 6 prospective, it would be a retrospective analysis 7 of previously gathered information such as what 8 you did in preparation for Doctor Beasley's 9 article plus something along the lines of the 10 second option they discussed here, a prospective 11 double-blind parallel comparitor study? 12 A. No, I'm not saying that. 13 Q. Okay. Explain what you mean by 14 a prospective study in that context. 15 A. I don't know what they mean by 16 historical prospective, but from my limited 17 knowledge of epidemiology, you have a 18 retrospectively defined cohort of patients that 19 you then follow forward in time. 20 Q. Okay. 21 A. So you define them 22 retrospectively, the cohort, and you follow this 23 patient cohort prospectively in time. 24 Q. So you gather the people back Page 310 1 up and put them on another study? How would you 2 follow forward in time? 3 A. That's right, you put them on a 4 study. 5 Q. So it would be almost in 6 effect, a rechallenge type of thing? 7 A. I'm not sure rechallenge was 8 involved or not. 9 Q. But it would be taking patients 10 who had already been on previous studies and 11 putting them back into some sort of controlled 12 study? 13 A. I'm not sure if the patients 14 would have to be on previous studies prior to 15 going in this study, they would have to meet 16 certain criteria which may include being on a 17 study or not, I'm not sure. But the key idea is 18 you take a defined population of patients, based 19 on something that's happened in the past, 20 historically, and then follow them prospectively. 21 Q. What's the different between 22 that and the second option they talk about, the 23 large prospective double-blind parallel 24 comparison trial? I'm talking about the Page 311 1 prospective cohort. 2 A. That's right. The option two 3 that they mention here is a randomized control 4 trial where you take a certain patient population 5 and you randomize them either to the treatment 6 under investigation or a comparitor, which might 7 be placebo. 8 Q. So the difference between the 9 second option and the rechallenge option is that 10 in the rechallenge option the patient population 11 would have to, in this case, have suffered from 12 suicidal ideation at some point prior to being 13 placed on the rechallenge study; correct? 14 MR. MYERS: Let me object. Are you 15 moving now forward to the third option, when you 16 say the rechallenge? 17 MS. ZETTLER: Right. 18 MR. MYERS: The question before was 19 about the epidemiological -- 20 MS. ZETTLER: Yes, I'm trying to 21 clarify something. 22 A. If you're comparing options two 23 and three now -- 24 Q. The difference is that -- Page 312 1 A. The difference is that I 2 believe the rechallenge study would take patients 3 who had been treated with antidepressants and 4 then retreat them with antidepressants. With the 5 second option, I'm not sure if these patients had 6 been on prior antidepressants or not. 7 Q. So you're taking, like, for 8 example, a patient population suffering from 9 major depressive disorder and putting them in a 10 study that's structured to evaluate or look at 11 the incidence of suicidality within that 12 population? 13 MR. MYERS: Which option? 14 MS. ZETTLER: Two, prospective study. 15 A. Yes. 16 Q. The rechallenge study could be 17 either people who had been on antidepressants 18 before or people who had been on antidepressants 19 before and had suffered a certain adverse effect 20 or some event; correct? 21 A. Correct. The only difference 22 between the two is the patient population that is 23 being studied. 24 Q. Okay. So they don't talk about Page 313 1 in this exhibit a purely retrospective review of 2 data gathered through the clinical trials, such 3 as what was done with Doctor Beasley's article; 4 correct? 5 A. That's correct. 6 Q. From reading this exhibit, 7 would you agree that the third option, the 8 rechallenge study, was thought, at least 9 according to this exhibit, to be the best option 10 out of the three? 11 MR. MYERS: Are you asking him is that 12 what it says? 13 MS. ZETTLER: Yes. 14 A. The statement here, the last 15 paragraph of the second page, first sentence, a 16 rechallenge study, although still with some 17 limitations, is considered the best of the three 18 options. 19 Q. And you don't know whether 20 another rechallenge study was performed? 21 A. No, I do not. 22 Q. You know that they were in the 23 process of developing a protocol for a 24 rechallenge study; correct? Page 314 1 A. Yes, that's correct. 2 Q. Do you know if a protocol was 3 ever completed? 4 A. No, I do not. 5 Q. Can you tell me what the second 6 portion of Exhibit Number 15 is, starting with 7 the page where it says antidepressant randomized 8 rechallenge study at the top? 9 A. This is a protocol outline for 10 an antidepressant randomized rechallenge study. 11 Q. Okay. Was this -- is this the 12 outline for the protocol that we talked about a 13 few minutes ago in Exhibit 12? 14 A. It's an outline for their third 15 option of the rechallenge study. 16 Q. Third option as reflected in 17 Exhibit 15? 18 A. Yes. 19 Q. Can you look at the next page 20 where it's got like a little post-it tab with 21 some writing stuck to the page. Is that your 22 handwriting? 23 A. I don't believe so. 24 Q. Okay. Do you recognize the Page 315 1 second portion of Exhibit 15? 2 MR. MYERS: Starting where? 3 MS. ZETTLER: Starting with the page 4 antidepressant randomized restudy protocol 5 outline, Pz 1275 1942. 6 A. I don't remember this, I don't 7 recognize it, per se. 8 Q. Could you look at the last page 9 of the exhibit, Pz 1275 1946, do you recognize 10 that? 11 A. I don't remember for sure 12 whether I saw this or not, but I have seen 13 similar things to it. 14 Q. Can you tell me what this is? 15 A. This is an outline for the 16 rechallenge study showing the different treatment 17 groups that patients could be randomized to, and 18 it shows how long they would be followed, at the 19 bottom, over the weeks. 20 Q. So there would be a placebo 21 wash-out period of approximately two weeks; 22 correct? 23 A. A single-blind period with 24 placebo, yes, for two weeks. Page 316 1 Q. And then after that would be 2 another eight weeks on the -- 3 A. Double. 4 Q. -- drug or placebo, whatever 5 they were randomized to? 6 A. That's correct. 7 (PLAINTIFFS' EXHIBIT NO. 16 WAS 8 MARKED FOR IDENTIFICATION AND 9 RECEIVED IN EVIDENCE.) 10 MR. MYERS: Nancy, before you start 11 asking him questions, I'm not going to let him 12 answer any questions about these exhibits that 13 we've submitted initially, en camera, and which 14 we argued about at the motion hearing, because 15 there's not been a ruling on them. 16 MS. ZETTLER: Yes there has, Larry. 17 MR. MYERS: I don't believe there has 18 been and I don't think it would be appropriate to 19 question him about exhibits that were given to 20 you and given to the court to decide some 21 discovery issues on which he's not issued an 22 order. 23 MS. ZETTLER: That's fine, we'll just 24 bring him back after an order is issued and we'll Page 317 1 question him about that. 2 MR. MYERS: All I can tell you -- and 3 that's the reason, until there is an order, I'm 4 not going to let him testify about the exhibits 5 that were submitted to Judge Potter on August 6 21st. 7 MS. ZETTLER: Well, you agreed to let 8 us see them, you've not asked for them back. If 9 you want to mark on there somehow that it's 10 confidential, that's fine. Judge Potter has 11 ordered that under -- except for very limited 12 circumstances, unless it is directly related to 13 efficacy or some other indication, then it is 14 something that is discoverable and you are to go 15 to the commissioner with those examples that you 16 feel are crucial to trade secret issues with 17 regards to the company and get her ruling prior 18 to withholding something from us. 19 MR. MYERS: Well, I don't -- 20 MS. ZETTLER: With regards to the 21 twenty-three boxes of documents that you guys 22 claimed were confidential, Judge Potter ruled 23 that nine out of ten of these exhibits, of which 24 this was one, were not confidential and were not Page 318 1 to be withheld. So as far as these specific 2 documents are concerned, he did rule that these 3 were not the types of things that would be 4 considered confidential under Judge Eckert's 5 March 31st, 1991 order. 6 MR. MYERS: Just so it's clear, these 7 documents on August 21st, and by this I mean the 8 one you've marked as 16, were submitted to the 9 judge after a random sampling by both yourself 10 and the special master, and submitted to the 11 judge and the master or commissioner first before 12 you ever received them. They were not produced 13 to you by us, but were turned over to you through 14 the court for purposes of making an argument on 15 August 21st. My understanding of the reason -- I 16 let you make your position clear, so let me make 17 mine clear. The reason that we have had a 18 commissioner or master appointed is to resolve 19 these issues surrounding documents which have 20 been withheld, and, you know, this isn't a 21 document that we've produced to you for purposes 22 of discovery, but which was given to you by the 23 court for purposes of that argument. So, you 24 know, I'm not trying to give you a hard time, I'm Page 319 1 just telling you that until those issues are 2 resolved, I'm not going to permit this witness or 3 any other witness to testify about any of those 4 documents. Any of these other documents that you 5 have that we've produced to you for purposes of 6 deposition, by all means question him about them. 7 But there's no reason to get in a big 8 disagreement, we've done pretty well for a few 9 days this week. 10 MS. ZETTLER: I guess it depends on 11 what you consider done pretty well. This was 12 obviously something that Judge Potter had looked 13 at and had determined was not -- if anything, I 14 believe that he has already ruled that the 15 majority of the documents that you withheld, 16 based on the random sampling, were not the types 17 of documents that were to be considered trade 18 secret. That you, whether it was done 19 intentionally or zealously, in representation of 20 your client, interpreted Judge Eckert's order 21 extremely overbroadly and that ninety percent of 22 the documents you withheld were inappropriately 23 withheld according to the order and according to 24 the categories that you yourself placed on these Page 320 1 documents. You, meaning the defense, Lilly and 2 their defense counsel. 3 As far as these specific documents are 4 concerned, he ruled that ninety percent of those 5 documents, ninety percent of them, were not to be 6 withheld, and he was referring specifically to 7 the random sampling that he reviewed as part of 8 his ruling the other day, that these documents 9 are to be turned over unless there are specific 10 examples that you guys can pull that you have to 11 show the commissioner to make a determination as 12 to whether or not they in fact are trade secret 13 type documents, confidential information. 14 Now, if you're going to instruct him 15 not to answer questions based on these documents, 16 then I'm going to bring him back after that 17 ruling is clarified, if you feel it still needs 18 to be clarified, and I'm going to ask him 19 questions about this document. 20 MR. MYERS: Well -- 21 MS. ZETTLER: We could do that. But 22 given the fact that this is a listing of 23 publications, and only a listing of publications, 24 I don't see where you have any basis to object to Page 321 1 him answering any questions about it. He 2 testified yesterday that he has been the author 3 of or participated in the preparation of numerous 4 articles. There are specific articles, in 5 particular the akathisia article, that he went 6 through yesterday, that are directly relevant to 7 this case. I believe I can refresh his 8 recollection as to whether or not that article 9 was published by referring to this exhibit. 10 MR. MYERS: Well, let me say this: I 11 don't agree with, and I don't have to agree with 12 your characterization of what the judge has 13 concluded, ordered, not ordered or otherwise. 14 All I'm telling you is as relates to the 15 documents which were furnished at the August 21st 16 hearing, I'm not going to permit him to testify 17 about them. Now if you want to ask him questions 18 about specific publications which you may believe 19 from some source that he was the author about -- 20 MS. ZETTLER: Yes, his own testimony 21 yesterday, Larry. 22 MR. MYERS: I understand that he talked 23 about publications, no question about it. If you 24 want to ask him about specific publications, go Page 322 1 ahead, I don't have a problem with that. All I'm 2 saying is he's not going to testify about that 3 specific document -- 4 MS. ZETTLER: Then he's coming back. 5 MR. MYERS: -- or any other one. 6 MS. ZETTLER: Then he's coming back. 7 And I believe that Judge Potter's order, the 8 rulings the other day, whether clearly, clearly 9 related to these documents. And that -- I'm 10 going to ask for sanctions if you're not going to -- 11 I'm going to ask for my costs in coming down here 12 to take the deposition and reconvening his 13 deposition again. 14 MR. MYERS: If there's a specific 15 publication in there that you believe he was an 16 author on, why don't you just ask him about that 17 without attaching the document to the deposition, 18 without making it an exhibit, and without 19 examining him about it, and maybe we can cut 20 through it and get to where you want to go while 21 at the same time, at least for the time being, 22 satisfy my concern. I'm not trying to thwart you 23 from examining him about publications where he 24 was an author or even about prospective Page 323 1 publications where he was an author, but I'm not 2 going to let him testify about the document in 3 general. But if there's a publication on that 4 that he's an author of, ask him about that, and 5 maybe that's the way to resolve it. 6 Q. (BY MS. ZETTLER) Do you 7 remember a study being done on discriminate 8 validity of Hamilton rating scales for 9 depression? 10 MR. MYERS: A study? 11 MS. ZETTLER: Yes, a study. 12 A. I'm not sure. 13 Q. Do you remember -- I'm sorry, 14 you're not sure? 15 A. I'm not sure about this 16 particular study, I would need more explanation 17 of what it was. 18 Q. I have no idea, I'm trying to 19 figure this out. But until I can show you this 20 exhibit, Doctor, I guess I don't think either one 21 of us is going to get anywhere. It says the 22 objective, okay, to use the three thousand 23 sixty-five data base to determine a restructuring 24 of the Ham-D rating scales based on discriminate Page 324 1 function test on Tomoxetine data base. Does that 2 refresh your recollection? 3 A. Could you read that one more 4 time for me? 5 Q. Sure. To use the, quote, three 6 oh six five, unquote, data base to determine a 7 restructuring of the Ham-D rating scale based on 8 discriminate function tests on Tomoxetine data 9 base? 10 A. Yes, I do, in general, remember 11 that. 12 Q. Tell me about it. What was the 13 study about? 14 A. In concept, this was a 15 retrospective look at the Hamilton depression 16 scale to see if there was a way that depression 17 could be measured using the different items in 18 the Ham-D scale other than just summing them up, 19 which is the current practice. For example, do 20 some of the items weigh more than other items 21 within the scale or is the straight sum proper or 22 are there other items that reflect true changes 23 in depression more than others. Right now the 24 total is used, but there might be different items Page 325 1 within the Ham-D that might change more than 2 others. So this is, I believe, and I'm not sure 3 on the specifics, but I believe it's an attempt 4 to look in more detail at the Ham-D scale and 5 different items within that scale. 6 Q. Do you -- was that paper ever 7 published to your knowledge? 8 A. Not to my knowledge. 9 Q. Why not? 10 A. I don't know how far that 11 investigation went. 12 Q. Under the common section on 13 here it says: To achieve some confusing results, 14 trying to understand them, dated 4-1-91. Is that 15 your recollection of what happened? 16 A. I don't remember the outcome of 17 that, that particular investigation. 18 Q. Under co-authors it has Saylor, 19 Beasley, Enas and consultants. Do you remember 20 who the consultants were on that? 21 A. No, I do not. 22 Q. Were those statistical 23 consultants or medical consultants? 24 A. I don't know. Page 326 1 Q. Is Wiltse a Ph.D? 2 A. Yes, he is. 3 Q. Would Doctor Wiltse know 4 whether or not the confusion was ever resolved? 5 A. I'm not sure what that 6 confusion refers to. Doctor Wiltse, I believe, 7 may have been more involved with that study as 8 his role as a practicing statistician, but I'm 9 not sure how this particular investigation was 10 resolved. 11 MS. ZETTLER: So, Larry, are you going 12 to let me read off this thing, you're going to 13 trust me to read off it appropriately, but you're 14 not going to -- 15 MR. MYERS: I'm going to give you the 16 benefit of the data by asking him about 17 publications he's associated with, yes. 18 MS. ZETTLER: That's really kind of 19 you. 20 MR. MYERS: You're welcome. 21 Q. I think I asked you this 22 yesterday, if I did, please forgive me, but the 23 akathisia study, that paper that you worked on, 24 was that related to hostility? Page 327 1 A. No. And to clarify, I did not 2 work on a paper dealing just with akathisia, I 3 was involved with a paper looking at adverse 4 events and their relationship to suicidality. 5 Q. So it was suicidality as 6 opposed to homicidality? 7 A. That's correct. 8 Q. Did you ever work on projects 9 regarding homicidality and the use of Fluoxetine? 10 A. I reviewed some work there, but 11 I was not the statistician involved with these 12 particular analyses. 13 Q. Does the title Hostility in 14 Patients Taking Fluoxetine refresh your 15 recollection as to whether or not you reviewed 16 that paper? 17 A. I may have reviewed that paper 18 because I reviewed all the papers under 19 preparation during this time period. 20 Q. Do you recall a paper titled 21 Meta Analyses, The Association Between Drop-out 22 Rates and Ham-D Three Scores in Several Clinical 23 Trials Concerning Fluoxetine, Tricyclics and 24 Placebo? Page 328 1 A. I do remember the gist of that. 2 Q. Is that a different paper than 3 the Beasley BMJ? 4 A. Yes, it is. 5 Q. When you say you remember the 6 gist of it, what do you mean? 7 A. Well, I remember the concept, 8 looking at differential drop-out rates between 9 treatment groups. 10 Q. If I told you that you were 11 listed on this exhibit or piece of paper as the 12 lead person on that article, would that refresh 13 your recollection as to the details of that 14 article? 15 A. Again, I remember the concept 16 of looking at differential drop-out rates, across 17 treatment groups. 18 Q. What happened, was that article 19 published? 20 A. No. 21 Q. Why not? 22 A. The work did not result in any 23 publication or I'm not even sure how far the 24 analysis went on that. Page 329 1 Q. Why didn't it result in 2 publication? 3 A. It was never submitted nor was 4 the work ever completed, possibly, I don't 5 remember the details. 6 Q. Why wasn't it submitted? 7 A. Because the work was never 8 completed. I don't remember the details of how 9 far the work was, but it was not completed. 10 Q. Did you decide to discontinue 11 work on the paper? 12 MR. MYERS: You, as in him personally? 13 MS. ZETTLER: Yes. 14 A. Not me personally. It's a 15 research interest item of mine, and if I had time 16 I would continue to pursue all kinds of things, 17 including this concept. 18 Q. Was it your work that was 19 discontinued on the paper or was work on the 20 paper generally discontinued? 21 A. No, it was in general, not my 22 work alone. 23 Q. Do you recall why it was 24 discontinued in general? Page 330 1 A. I can only speak to my work, I 2 cannot speak to in general. 3 Q. Okay. 4 A. My work was discontinued 5 because in a finite period of time I could only 6 work on so many projects, and so this one was one 7 that I didn't pursue relative to other projects. 8 Q. How about Doctor Beasley, did 9 he work on that article? 10 A. I'm not sure to what extent 11 Doctor Beasley worked on the article. Many 12 publications are at different stages of 13 development, some are in concept stages, some are 14 where you get a data base, some where you 15 actually analyze the data base, and some where 16 you're writing it up. So Doctor Beasley may have 17 been involved only at the concept stage or 18 preconcept, I'm not sure what stage he was 19 involved with. 20 Q. Did he transfer the project to 21 another person in your department? 22 A. No, not fully, because this is 23 a personal research interest of mine and so I 24 wouldn't, for this item, just say here, I'm Page 331 1 divorcing myself from it, you take it over. 2 Q. Was this article your idea? 3 A. I wouldn't say fully, no. 4 Q. Okay. I'm trying to understand 5 why work was stopped on the article generally, 6 because I'm sure everybody didn't stop the 7 project just because you didn't have time to work 8 on it; correct? 9 A. Yes, in general it was stopped 10 because there's many articles and publications 11 people are working on, and this one was not 12 resourced at the level the other projects were 13 because of time constraints. 14 Q. So are there plans to start 15 working on this article again? 16 A. There are no explicit plans to 17 do so. However, as I mentioned, this is a 18 personal research item that I would love to carry 19 out in the future. 20 Q. Okay. Do you know of anybody 21 else who was working on that article while you 22 were working on it? 23 A. I do not remember any 24 colleagues, in fact the statistical staff, we Page 332 1 have talked about the concept amongst a number of 2 individuals, but I don't remember if anybody said 3 yes, I am working on this particular project. 4 Q. Well if you're this co-author, 5 Sampson, Beasley, Faries and Bosomworth, does 6 that refresh your recollection? 7 A. Those are among the folks that 8 have discussed this concept, yes. 9 Q. How much work did you actually 10 do on the concept? 11 A. To the best of my knowledge, 12 all the work that was done on the concept was to 13 acknowledge that it would be interesting to 14 compare these drop-out rates. 15 Q. With Ham-D three scores? 16 A. And I'm not sure how far it 17 went beyond that. 18 Q. And the Ham-D three is the 19 suicidality question; correct? 20 A. Yes. 21 Q. Was this to look at the 22 thirty-five -- or three thousand sixty-five 23 patients to see how many of them had dropped out, 24 suicidal patients, how many had dropped out of Page 333 1 the study as opposed to how many completed the 2 study? 3 A. It's related to the thirty 4 sixty-five data base, and looking at when 5 patients drop out. Some patients drop out early 6 in the study, some patients are retained in the 7 study and drop out later. 8 Q. And you wanted to see if there 9 was a correlation between the patient dropping 10 out and a suicidal attempt or suicidal ideation 11 increase? 12 A. That's included in it, but it's 13 not the total picture. It's just comparing the 14 groups because they have differential treatment 15 drop-out rates. It might be, for example, that 16 patients drop out earlier in one treatment than 17 another treatment. 18 Q. Have you ever heard of meta 19 analysis two A through two D? 20 A. That rings a bell, yes. 21 Q. Can you tell me about that? 22 A. I do not remember the 23 particulars, I remember that two A through two D, 24 though, but I don't remember the particulars. Page 334 1 Q. Was that four different 2 studies, four different meta analyses? 3 A. It might have something to do 4 with the U.S. plus international -- you know, the 5 two different patient populations, I'm not sure 6 exactly, though. 7 Q. If I were to tell you this 8 document was dated April 22, 1991, would that 9 help you to remember what the two A through two D 10 would be? 11 A. No, the data alone would not 12 help me. 13 MS. ZETTLER: Until we have a ruling 14 from the judge, Doctor, I can't ask you anymore 15 questions about this. 16 MR. MYERS: Can we take a short break? 17 MS. ZETTLER: Sure. 18 (A SHORT RECESS WAS TAKEN.) 19 MS. ZETTLER: So to clarify this 20 record, Larry, you're objecting to me marking 21 what was originally marked as Exhibit 16 as an 22 exhibit and attachment to this deposition? 23 MR. MYERS: Yes. 24 MS. ZETTLER: Okay. For the record, Page 335 1 Enas Exhibit 16 originally was Exhibit 16, a 2 document in which the first page is marked Pz 3 1274 881, running for a change straight through 4 to Pz 1274 898. First page is dated April 22, 5 1991. There's a rather lengthy list of 6 recipients including people who are carbon 7 copied, and the cover page says re, quote, 8 status, dash, funds for publications Fluoxetine, 9 unquote, dash, April 22, 1991. 10 Q. Doctor, were you asked at some 11 point to start turning your Fluoxetine related 12 documents over to the legal department at Lilly? 13 A. Yes. 14 Q. Do you remember when they began 15 asking you to turn those documents over? 16 A. I don't remember. 17 Q. Was it before 1990? 18 A. I can't remember the date. 19 Q. Can you give me a year? 20 A. No, I cannot. 21 Q. How often have you been asked 22 to turn over documents to the legal department, 23 documents related to Fluoxetine? 24 A. I do know that periodically Page 336 1 there's a notice sent out to bring your documents 2 that you turn in current, turn them in to the 3 legal department. I don't know how often that 4 is, but it is periodic. 5 Q. Okay. Could it be quarterly? 6 A. It could be. 7 Q. Do you store or have you in the 8 past stored many of your documents on floppy 9 disks? 10 A. No. 11 Q. Do you store any documents on 12 floppy disks? 13 A. I don't currently store any 14 documents on floppy disks. 15 Q. Have you ever stored documents 16 on floppy? 17 A. I believe I have. 18 Q. Okay. Do you recall ever 19 copying any of your floppy disks containing 20 Fluoxetine information and turning them over to 21 the legal department? 22 A. I haven't copied things on 23 disks and turned them over because all of my 24 documents, my secretary has to, and they've Page 337 1 always been turned over and updated. 2 Q. Do you have recollection of 3 copying information that is on a floppy disk onto 4 another disk and turning that disk over to the 5 legal department? 6 A. I don't have recollection of 7 that. 8 Q. Do you have a recollection of 9 not doing that? 10 A. No. 11 Q. What's a program listing? 12 A. In the statistical department, 13 a program listing might be an SAS program that 14 does a particular analysis on a data base. 15 Q. What's a program test listing? 16 A. That's a term I'm more 17 unfamiliar with, so I'm not certain what that is. 18 Q. How about a transfer 19 documentation binder? 20 MR. MYERS: What was that, I'm sorry? 21 MS. ZETTLER: Transfer documentation 22 binder. 23 A. I'm not certain what that is. 24 Q. A version to the standard Page 338 1 binder? 2 A. I don't know what that entails, 3 it sounds like a systems binder. 4 Q. DE binder? 5 A. I don't know what that is. 6 Q. Data entry? 7 MR. MYERS: What is data entry, is that 8 the question? 9 Q. No. Could it be data entry 10 binder? 11 A. It could be, but I'm not 12 certain, we don't use that in our department. 13 Q. Are you aware of any Lilly 14 employees who worked on Fluoxetine storing 15 information on floppy disks? 16 A. No, I am not. 17 Q. Have you heard of floppy disks 18 being transferred to the legal department? 19 A. I have not heard of floppy 20 disks being transferred to the legal department 21 explicitly. 22 Q. Have you ever filled out a form 23 listing what was being sent to legal? 24 A. I can't remember. Page 339 1 Q. How about your secretary, does 2 she fill out forms as to which of your documents 3 are being sent to legal? 4 A. I don't know if she fills out a 5 form or not, I do know that my secretary sends 6 the documents to the legal department. 7 Q. She's the person who collects 8 the documents and transfers them to legal, your 9 documents? 10 A. No, I collect the documents and 11 put them all together, I keep a current file of 12 anything that relates to Prozac, and 13 periodically, when the request is made, she then 14 takes that and sends it to legal. 15 (PLAINTIFFS' EXHIBIT NO. 17 WAS 16 MARKED FOR IDENTIFICATION AND 17 RECEIVED IN EVIDENCE.) 18 Q. Have you had a chance to review 19 Exhibit 17, Doctor? 20 A. Yes. 21 Q. Does this refresh your 22 recollection as to the document collection 23 process and the details of it? 24 A. It doesn't refresh my memory to Page 340 1 all the details, like how often these requests 2 were sent out, but yes, it's part of that 3 retrieval process. 4 Q. Does this refresh your 5 recollection as to whether or not you filled out 6 a form or were asked to fill out a form listing 7 what you were sending to legal? 8 A. It doesn't help me recollect 9 whether I filled out a form or not, but it does 10 mention that forms list things that are sent to 11 legal, I'm not sure if I filled those out or my 12 secretary filled them out. 13 Q. Do you have a file that you 14 keep of the lists of what you send to legal, 15 Fluoxetine related documents that you sent to the 16 legal department? 17 A. I don't personally have a file. 18 I keep a file of anything related to Fluoxetine, 19 and then periodically, as I said, they send that 20 out to the legal department. 21 Q. Does your secretary keep a list 22 of what's been sent to the legal department? 23 A. I'm not sure. 24 (PLAINTIFFS' EXHIBIT NO. 18 WAS Page 341 1 MARKED FOR IDENTIFICATION AND 2 RECEIVED IN EVIDENCE.) 3 Q. Have you had a chance to look 4 at Exhibit 18? 5 A. Yes. 6 Q. Do you recognize that exhibit? 7 A. No. 8 Q. You don't. Do you have any 9 idea why it would have been produced as being in 10 your file? 11 A. I inherited many things, again, 12 from previous statisticians. 13 Q. Are you familiar with the 14 subject matter of the exhibit? 15 A. I'm not familiar with this 16 study at all. 17 Q. Does this exhibit, in the state 18 that it's in, with the information marked out, 19 give you any idea of who worked on this study 20 from Lilly? 21 A. No. To the best of my 22 knowledge, I can't tell. 23 Q. Does it -- can you tell who 24 worked on this study at all from this exhibit Page 342 1 with the information removed as it is? 2 A. No, I cannot. 3 Q. From your review of the 4 exhibit, and if you need more time to take a 5 closer look at it, that's fine, but is this 6 similar to what we talked about when we talked 7 about the prospective study earlier that was 8 discussed in, I believe it was, Exhibit 15? 9 MR. MYERS: Is the question is that -- 10 MS. ZETTLER: The type of study that 11 was discussed. 12 Q. Is it the type of study that 13 was discussed in Exhibit 15, alternative two? 14 A. By alternative two, do you mean 15 the large prospective double-blind parallel 16 comparison trial? 17 Q. Right. 18 A. No. 19 Q. Does the Exhibit 18 reflect, at 20 least in structure from the exhibit, any of the 21 possible alternatives listed in Exhibit 15? 22 MR. MYERS: You mean can this fit under 23 one of those three headings or descriptions? 24 MS. ZETTLER: Yes. Page 343 1 A. I'm not sure whether it could 2 or not. 3 Q. Do you know if this study was 4 ever performed? 5 A. I do not know. 6 (PLAINTIFFS' EXHIBIT NO. 19 WAS 7 MARKED FOR IDENTIFICATION AND 8 RECEIVED IN EVIDENCE.) 9 Q. Have you had a chance to look 10 at Exhibit 19? 11 A. Yes. 12 Q. Is Exhibit 19 related to 13 Exhibit 18? 14 A. Yes. 15 Q. How can you tell? 16 A. The global project tracking 17 number on Exhibit 18 matches the global project 18 tracking number on Exhibit 19. 19 Q. Okay. Does Exhibit 19 indicate 20 whether or not the study talked about in Exhibit 21 18 was actually performed? 22 A. Would you repeat that? 23 Q. Sure. Does Exhibit 19 indicate 24 whether or not the study that is discussed in Page 344 1 Exhibit 18 was actually performed? 2 A. Yes. 3 Q. And was it performed? 4 A. Yes, this Exhibit 19 is a study -- 5 a report of that study. 6 Q. Okay. Is this a report that 7 was submitted to the FDA to your knowledge? 8 A. I don't know. 9 Q. Is there any way to tell from 10 the exhibit whether or not it was submitted to 11 the FDA? 12 A. I cannot tell personally from 13 this exhibit, but I don't know what marks or 14 documentations show that a document is actually 15 submitted to the FDA, so I can't determine that. 16 Q. Were you aware of this study 17 before today? 18 A. No. 19 Q. Were you made aware of this 20 study with regards to any of your work on the 21 Beasley article? 22 A. I don't believe so. 23 Q. Were you made aware of this 24 study with regards to any of your work on the Page 345 1 submissions made to the advisory committee? 2 A. I can't remember. 3 Q. Well, obviously if you weren't 4 aware of this study before today, you probably 5 weren't, right? 6 A. I can't remember if somebody 7 made me aware or not, I may have forgotten. But 8 when I saw it today, it looked like I had seen it 9 for the first time. 10 Q. Does the statistical analyses 11 in Exhibit 19 differ in any way from the 12 statistical analyses performed in regards to the 13 Beasley article? 14 A. Yes. 15 Q. In what way do they differ? 16 A. The study designs are 17 different. The Beasley article dealt with 18 comparing treatment groups to treatment groups in 19 a randomized setting. This Exhibit 19 does not 20 reflect a randomized setting, but an open label 21 single treatment group study. So the methods 22 would naturally differ because it's a different 23 study design. 24 Q. Were you able to find what the Page 346 1 results of this study reflected in 19 were? 2 A. I'm sorry? 3 Q. Can you find in there for me 4 where the results of the study reflected -- the 5 results of the study in 19? 6 A. There's a conclusion that's 7 mentioned on page thirty-four that says Prozac 8 twenty was found to be significantly effective in 9 relieving depression and suicidal thoughts among 10 the participants of the trial. 11 Q. Page twenty-four? 12 A. Thirty-four. 13 Q. Prozac twenty was found to be 14 significantly effective in relieving depression 15 and suicidal thoughts among the participants in 16 the trial? 17 A. Correct. 18 Q. Did they mean the actual 19 patients who took the drug, when they say 20 participants, are they talking about doctors and 21 statisticians and consultants, et cetera, who may 22 or may not have been involved? 23 MR. MYERS: Are you asking him what 24 participants, who he interprets that to mean? Page 347 1 MS. ZETTLER: Right. 2 A. I interpret participants to 3 mean the actual patients who were studied in the 4 trial. 5 Q. Does this report reflect 6 whether or not the patients that participated in 7 the trial were interviewed and whether those 8 interviews are reflected in this study, in this 9 report? 10 A. As a statistician, I can only 11 comment on statistical questions, and that 12 question is not statistical, it has to do with 13 the medical investigator interviewing patients to 14 come up with the MADRS score, for example, which 15 is used as one of the efficacy criteria here. So 16 I assume that the physicians collected those data 17 appropriately, the Ham-D's and MADRSs, et cetera. 18 Q. So they were able to 19 extrapolate the results of the tests submitted by 20 the doctors to say that the patients themselves 21 felt that they got better? 22 MR. MYERS: They, who is they? 23 MS. ZETTLER: Lilly. 24 Q. In other words, whoever wrote Page 348 1 this report and came to the conclusion on page 2 thirty-four that Prozac was found to be 3 significantly effective in relieving depression 4 and suicidal thoughts among the participants in 5 the trial, came to that conclusion based on the 6 testing that was -- that they were submitted to, 7 the patients were submitted to by the 8 investigators. 9 A. That's my understanding as I 10 read this, that the investigators collected the 11 data from the patients and that that data was 12 then analyzed by the person who analyzed this 13 data and produced these graphs and things like 14 that, and then the authors came up with this 15 conclusion based on those results. 16 Q. Would you look at page five, 17 please? 18 A. Okay. 19 Q. At the bottom of the page under 20 study procedure schedule, it says, in a graph, 21 history/examination, MADRS total, Ham-D, CGI, ST, 22 PS, pill count and adverse events; correct? 23 A. Yes. 24 Q. Okay. Do you know if the MADRS Page 349 1 total -- if the MADRS test is something that the 2 patient fills out or the investigator fills out 3 in an interview with the patient? 4 A. I'm not certain. 5 Q. How about the Ham-D, is that 6 something that the patient fills out or something 7 that the investigator fills out? 8 A. To the best of my knowledge, 9 that's something the physician or investigator 10 fills out. 11 Q. Are you familiar with the 12 clinical global impression scale? 13 A. I've heard of it. 14 Q. Is that something that the 15 investigator fills out or the patient to your 16 knowledge? 17 A. I assume, because it's a 18 clinical global impression score, that the 19 clinician fills it out. 20 Q. What's the ST, suicide thought, 21 have you ever heard of that before? 22 A. I don't believe I have heard it 23 before in this vein. 24 Q. Other than somebody having a Page 350 1 suicidal thought, that's the vein you're thinking 2 of? 3 A. Correct. 4 Q. Do you know if there was a test -- 5 can you tell from this exhibit if there was a 6 test that was administered to rate suicidal 7 thoughts? 8 A. I've noticed that on page 9 eight, I assume -- that may not be page eight, 10 there's no page number, eleven point two, rating 11 scales. 12 Q. Okay. 13 A. That last paragraph says: On 14 the occasion of each visit, each patient's 15 suicidal thought score, parentheses, ST, was 16 extracted from the MADRS item number ten and 17 reported separately. 18 Q. So that was a determination, at 19 least it appears from this paragraph that that 20 was a determination made by reviewing MADRS item 21 number ten? 22 A. That's how I interpret that 23 section there. 24 Q. Do you have any other Page 351 1 recollection or any recollection of anything else 2 related to 19? 3 A. No, I do not. 4 Q. Who developed the methodology 5 that was used in support of the Beasley BMJ 6 article? 7 MR. MYERS: Which methodology? 8 MS. ZETTLER: Any of this, the 9 statistical analyses used, who determined what 10 data bases were going to be used, things of that 11 nature. 12 A. I can speak most to the 13 statistical methodology that was used there. 14 Q. Okay. 15 A. I'm not certain of the 16 originators of the methodology, but our 17 statistical staff in conjunction with our 18 consultant come through many ways to analyze 19 data, because there are many, and came up with 20 what they felt would be the soundest methodology 21 for these kinds of data, using published 22 references and the like. 23 Q. Were there a number of 24 different types of statistical analyses apart Page 352 1 from the data or used with the data or was it 2 just that one? 3 A. There were different types of 4 analyses applied to the data. 5 Q. Okay. Did the results differ 6 depending on the analysis that was applied? 7 A. I can't remember exactly, but I 8 believe substantively they did not differ. 9 Q. Why was it decided to use the 10 particular analysis that was ultimately used in 11 the BMJ article? 12 A. For one, it used stratification 13 that we discussed, and it was felt that 14 stratification for the prime analysis was 15 appropriate. So that's the methodology that 16 incorporates stratification. Now within 17 stratified analyses, there's a number of analyses 18 that you can choose, and this particular one was 19 chosen because of some issues that I'm not fully 20 versed on now, but it's appropriate for the data 21 at hand. 22 Q. Was stratification used prior 23 to August, 1991? 24 MR. MYERS: By Lilly? Page 353 1 MS. ZETTLER: Yes. 2 MR. MYERS: In connection with what? 3 MS. ZETTLER: The BMJ article. 4 A. I can't comment on the exact 5 date, but to my memory stratification was first 6 used in a report in a submission to the FDA to 7 Dr. Paul Liebert of the suicidality data. 8 Q. The suicidality data used in 9 support of the Beasley article? 10 A. I'm not sure if it was exactly 11 the same data, but it was the report issued to 12 the FDA based on his request to do what's called 13 a Mantel Hansel test, which is the stratified 14 test that we did. And we used odds ratios in 15 that report to the FDA, and in the BMJ paper we 16 used what is called the incidence difference, 17 using the Mantel Hansel stratification. 18 Q. Dr. Liebert asked specifically 19 that you run the Mantel Hansel analysis on 20 suicidality data? 21 A. I forget the specific request, 22 but in concept I think that was the gist of it. 23 Q. Do you know if this was before 24 or after the advisory committee meeting in Page 354 1 September of 1991? 2 A. I believe it was before. 3 Q. Do you know how long before? 4 A. I do not remember. 5 Q. More than a year? 6 A. I can't remember, it may have 7 been a year. 8 Q. Okay. Yesterday we were 9 talking about the spontaneous reporting system 10 with the FDA where the post-marketing spontaneous 11 events are collected. Do you remember that? 12 A. I remember our discussion. 13 Q. And Paul asked you whether or 14 not a statistical analysis of that data was done 15 and you said that to your knowledge no; correct? 16 A. That's correct. 17 Q. Okay. And part of the reason 18 you said that was not done was because it would 19 be hard to do that without a comparitor, there's 20 no comparitor in the SRS, the spontaneous 21 reporting system; correct? 22 A. That's right, that's one 23 argument. 24 Q. Any other arguments why you Page 355 1 wouldn't do that? 2 A. I believe some were mentioned 3 in one of the exhibits here with respect to 4 post-marketing reports, but the biggest argument 5 is that when you -- post-marketing reports come 6 from a variety of patients, and you come up with 7 some kind of rate, patients were treated for 8 different lengths of time and then you don't know 9 what that means because you don't have a group 10 that was untreated at the same time. So the 11 biggest argument is the lack of comparitor from a 12 statistical point of view. There might be other 13 reasons that a physician might give, but from a 14 statistical point of view it's the comparison 15 issue. 16 Q. Are you aware that ninety 17 percent of the adverse reactions or adverse 18 events, post-marketing, are reported to the FDA 19 by the manufacturer of the drug? 20 MR. MYERS: Before he answers, let me 21 object to the form because I think it assumes 22 facts not in evidence and I don't know whether 23 your question is restricted to this drug or drugs 24 generally. Page 356 1 MS. ZETTLER: Drugs generally. 2 A. I'm not aware of the actual 3 percentage. 4 Q. Okay. Are you aware that most 5 adverse event reports made to the FDA after a 6 drug is marketed are made by the manufacturer of 7 the drug? 8 A. In general, I would suspect 9 that to be true, but I can't be certain. 10 Q. Why couldn't a comparison of 11 spontaneous adverse event reports made with 12 regards to Prozac be done with spontaneous 13 adverse event reports made with regards to other 14 antidepressants, why couldn't you use those two 15 groups to analyze? 16 A. Well, again, it goes back to 17 assuring the comparability of those two patient 18 groups. There is an issue here of time. Those 19 antidepressants, for example, may be introduced 20 to the market at vastly different times, so 21 there's a time differential. Whereas the patient 22 characteristically -- a certain group of patients 23 are being treated with one antidepressant, and 24 completely different kind of patients are being Page 357 1 treated with Prozac. So again, there's many 2 factors that might mitigate against just blindly 3 comparing those two together. 4 Q. Couldn't you find out or make a 5 reasonable estimate of how many people had been 6 prescribed, say, Imiprimine and a tricyclic 7 antidepressant with the number of people who have 8 been prescribed Prozac and do an analysis of the 9 percentages of adverse events that were reported 10 with regards to each and percentages of 11 suicidality adverse events that were reported? 12 A. Yes, you can compare the two 13 numbers, but again, I would qualify that by 14 questioning how certain one would be that those 15 two numbers came from the same kind of patients, 16 the Imiprimine patients might be different in 17 substantial degree to the Prozac treated 18 patients. 19 Q. You mean different in what 20 disease process they're suffering? 21 A. That's, again, a medical 22 question, I can't answer that. But right, there 23 might be some biological or medical reasons. 24 Q. If the majority of adverse Page 358 1 events that are reported to the FDA are reported 2 by a manufacturer of a drug, and I'm not limiting 3 it just to Prozac, okay, isn't it fair to say 4 that the manufacturer would have information 5 regarding the adverse events it reports? In 6 other words, for what indication a drug was 7 given, how long a person had been on the drug, 8 things of that nature. 9 A. I'm not certain how much the 10 manufacturer knows about those particular 11 patients. 12 Q. Have you ever seen a sixteen 13 thirty-nine form? 14 A. I've seen -- I remember a blank 15 sixteen thirty-nine form. 16 Q. Okay. Are you aware that on 17 the sixteen thirty-nine there's a space to write 18 down what indications the drug was used for? 19 A. I'm not familiar with that 20 space, but I'm not surprised that it would be 21 there, it seems reasonable. 22 Q. It also asks for information 23 with regards to the duration of therapy on a 24 particular drug? Page 359 1 A. Okay. I can't remember. 2 Q. Okay. It also asks for 3 demographic data such as age and sex of the 4 patient. 5 MR. MYERS: Are you telling him or 6 asking him? 7 Q. I'm asking him. If you know. 8 A. I'm not certain. Again, I've 9 seen a blank sixteen thirty-nine, but I don't 10 remember the individual elements. 11 Q. Assuming all that information 12 is there, plus information or at least a space to 13 list information about concommitant and 14 medications and things of that nature, do you 15 think that you could narrow down the field to 16 where you could get to a fairly reasonable 17 comparitor group? In other words, can you find 18 similar enough patients to be able to run some 19 sort of statistical analysis? 20 A. Again, I would have to defer to 21 my medical colleagues because I'm not certain of 22 all the variables that are important in matching 23 patients, and the sixteen thirty-nine may not 24 contain all the information that's essential to Page 360 1 making a clean match. 2 Q. Are you aware that the number 3 of spontaneous adverse events reported with 4 regards to Fluoxetine are at least 5 percentage-wise or number-wise significantly 6 greater than other antidepressants? 7 A. I've heard that, but I'm not 8 familiar with the actual rates. 9 Q. Could that be one reason why a 10 statistical analysis was never done to compare? 11 A. No. 12 Q. Why not? 13 A. If people wanted to compare 14 rates, they're right there. I don't know what 15 the rates are, but you have rates for different 16 comparisons, then the person has obviously made a 17 comparison right then and there, you don't need 18 fancy statistics. 19 Q. So you can look at the -- I'm 20 using this for an example: Say there are ten 21 thousand adverse suicidality related adverse 22 events reported for Fluoxetine, and there's only 23 a thousand reported for, say, Imiprimine, you can 24 see the difference just by looking at those Page 361 1 numbers? 2 A. What I'm saying is somebody can 3 compare those, but I'm not certain whether or not 4 it would be a valid comparison, because it might 5 be comparing patients, different lengh of 6 exposure, whatever it might be, so just comparing 7 the numbers might be misleading. 8 Q. So you can explain away those 9 differences using statistics; correct? 10 MR. MYERS: What differences? 11 MS. ZETTLER: Differences in the number 12 of adverse events. 13 A. Not. It's not statistical 14 reasoning, again, it's a medical reason and, for 15 example, we know that with a media, for example, 16 and the proliferation of adverse events as things 17 get out in the media. That's another thing that 18 might be different for one drug than another 19 drug. So what I'm saying is there's lots of 20 reasons that are not statistical, but lots of 21 phenomena that might explain why you see 22 differences between treatments more than just the 23 statistical process. 24 Q. So when you refer to the media, Page 362 1 you're saying because there may be some adverse 2 events that are reported in the media, it may 3 cause other reports that may or may not have been 4 reported and got media attention? 5 A. That's what I've heard. I'm 6 not an expert in that area, but that's what I've 7 heard. 8 Q. Have you also heard that 9 according to the FDA only about one in ten 10 adverse events are actually reported? 11 A. No, I'm not familiar with that 12 rate. 13 Q. Well if that is true, couldn't 14 it be said that the media attention was doing 15 something positive in that it may have prompted 16 people who would not have reported adverse events 17 to report them but for the media? 18 A. Possibly so. But again, if 19 you're asking about proper comparisons, you want 20 everything to be equal across all patients and 21 all treatments in order to make a clean 22 comparison. 23 Q. Just because an adverse event 24 is reported because of media attention doesn't Page 363 1 mean that that person isn't similar to other 2 people on the drug, does it? 3 A. No, that's correct, not 4 necessarily. 5 Q. So you could take the numbers 6 and take the individual cases and make that -- 7 try to make a determination as to whether or not 8 they're reasonably similar with other patients 9 who reported adverse events; correct? 10 A. That's possible. But as a 11 statistician, what spontaneous reports can do is, 12 you know, it might signal something that you need 13 to go back and in your best data base, which 14 might be a randomized control data base, look for 15 that, because then in the randomized data base 16 you do have -- you're assured of comparability 17 between the treatment groups. 18 Q. This group of adverse events 19 regarding suicidality could be a pool from which 20 you could draw possibile patients for a 21 rechallenge protocol, isn't that true? 22 A. I'm not sure. Again, that's a 23 medical determination to include patients in a 24 trial, so I'm not sure if that would be the Page 364 1 appropriate group or not. 2 Q. Have you ever heard of PIEFRS, 3 P-I-E-F-R-S? 4 A. I've heard of them, yes. 5 Q. What is that, what are they? 6 A. I'm not certain other than I 7 have heard them linked with the FDA in some kind 8 of reporting. 9 Q. Possible increased event 10 frequency report? 11 A. Yes. 12 Q. Does that sound familiar? 13 A. Yes. 14 Q. Have you ever worked on 15 submission by Eli Lilly to the FDA regarding 16 possible increased event reports with 17 suicidality? 18 A. What do you mean by work on? 19 Q. Have you done an analysis or 20 have you gathered data or helped in any way to -- 21 A. Not that I'm aware of. 22 Q. Have you worked on any 23 protocols that have been protocols for clinical 24 trials that have been performed outside the Page 365 1 United States? 2 MR. MYERS: Fluoxetine? 3 MS. ZETTLER: Right. 4 A. I'm not certain whether I have 5 been intimately involved in the preparation of 6 such a document, I may have reviewed such a 7 document, but I'm not certain whether I have 8 written protocols for studies outside the United 9 States. 10 Q. Have you done any analysis of 11 data that was gathered through a study that was 12 performed outside the United States? 13 A. A Fluoxetine study? 14 Q. Right. 15 A. I'm not certain. 16 Q. Are you familiar with any 17 clinical trials that have been performed in 18 Taiwan? 19 A. Fluoxetine clinical trials? 20 Q. Right. 21 A. Performed in Taiwan. I'm not 22 familiar with any clinical trials that have been 23 performed in Taiwan. 24 Q. Going back to the Beasley Page 366 1 article in the seventeen U.S. double-blind 2 controlled studies, okay, when you said your 3 staff reviewed various clinical trials to see if 4 they should be included, did you mean that they 5 were given a list of clinical trials that they 6 could add to the seventeen or were they looking 7 at the seventeen to see if a particular trial 8 should be excluded from that number? 9 A. No. What I had said is that 10 our staff was involved in stating yes, we want to 11 look at all the randomized controlled clinical 12 trials, and that was the broadest definition by 13 which studies were included in that pool. 14 Q. And from that list, the 15 seventeen were picked or was the list that they 16 were given in response to that request the 17 seventeen? 18 A. Would you repeat that again, 19 the two options? 20 Q. Sure. I'm trying to figure out 21 if there was a list of, say, just for example, 22 fifty double-blind controlled studies, and from 23 that fifty, the group that was analyzed was 24 narrowed down to seventeen. Or was it a matter Page 367 1 of the people in your department or on your staff 2 looking at the seventeen controlled trials to 3 make sure all of those seventeen should be 4 included? 5 A. I'm not certain about the exact 6 process, but what I believe happened is our 7 staff, in conjunction with that, it's not just 8 the statistical -- 9 Q. Right. 10 A. -- it's medical, et cetera, 11 together said we want to look at all randomized 12 controlled clinical trials, now go find them and 13 let's look at them. 14 Q. Do you know -- but do you know 15 what happened after that, if it was narrowed down 16 to seventeen or if that seventeen -- were there 17 only seventeen controlled clinical trials? 18 A. I'm not certain. 19 Q. Do you have an idea? 20 A. No. To the best of my 21 knowledge, based on the determination that they 22 wanted to randomize trials, whether it be placebo 23 controlled or tricyclic controlled, and to the 24 best of my knowledge all such trials were Page 368 1 included, but I'm not certain. 2 Q. Did you work on the analysis of 3 data for submission to the advisory committee? 4 We talked so much about the Beasley article, I 5 can't remember if you said you had worked on it 6 or had not. 7 A. I analyzed data in preparation 8 for the advisory committee, yes, I was part of 9 the staff that did that. 10 Q. All of the studies that were 11 analyzed for submission to the advisory committee 12 were double-blind controlled studies, may have 13 been done in the United States or outside the 14 United States, or may have been involved with 15 other indications, but they were all double-blind 16 controlled studies; correct? 17 A. I'm not certain. I know that 18 the randomized controlled trials were part of the 19 submission, I'm not sure what else additionally 20 was a part of that submission. 21 Q. Okay. Your analysis was 22 limited to double-blind controlled studies? 23 A. My staff's analysis? 24 Q. Yes. Page 369 1 A. I believe for the most part it 2 was limited to the randomized controlled 3 double-blind studies. 4 Q. Was the same statistical 5 analysis applied to each group of data, in other 6 words the seventeen U.S. studies, the outside the 7 U.S. studies, the bulimia studies, the OCD 8 studies, was the same statistical analysis 9 applied across the board? 10 A. To the best of my knowledge, 11 the Mantel Hansel incidence difference analysis 12 was applied to each group of studies. 13 Q. The same P value was given, the 14 .05? 15 A. The same level of significance, 16 .05, I believe, was used in all of those 17 analyses. 18 Q. Was the same statistical 19 analysis applied to the data that you analyzed 20 for the akathisia article? 21 A. To clarify, that was not an 22 akathisia article. 23 Q. Since you can't remember -- I 24 know, I'm trying to be difficult, but because you Page 370 1 can't remember the title and -- 2 A. Okay, yes. 3 Q. I'm using that as a shortened 4 form. 5 A. Yes, the Mantel -- I believe 6 the Mantel Hansel incidence difference was 7 applied there as well. 8 Q. Are you familiar with the 9 Prozac conversion to standards project? 10 A. I'm familiar with that term, I 11 don't know the details. 12 Q. Can you tell me generally what 13 the project was about? 14 A. As far as I understand it, 15 there are multiple and have been multiple Prozac 16 data bases, and standards, as I understand it, is 17 a standard data base that other data bases that 18 may not have been in that form were converted to 19 that form. 20 Q. So it was done for consistency? 21 A. Consistency and the ability to 22 pool all the data together and analyze it as one. 23 But I'm not an expert in the systems side of it, 24 but to the best of my ability, that's just a Page 371 1 general Gestalt of what's there. 2 Q. Do you know what changes were 3 made to the data base? In other words was it 4 changes in adverse events, was it changes in 5 reporting of psychiatric scales, things of that 6 nature? 7 A. I don't know, I'm not certain 8 of, quote, what these changes were. To me my 9 impression is you have an apple in this picnic 10 basket, you take the same apple and put the same 11 apple into a bucket so the apple remains the same 12 apple, it's just that what it's housed in is 13 different. 14 Q. Are you familiar that there was 15 at one time an adverse event dictionary used at 16 Lilly called the SSAI, Signs, Symptoms and 17 Illnesses? 18 A. I'm not familiar with that. 19 Q. Are you aware that at one point 20 Lilly changed from the ELECT dictionary to the 21 COSTART dictionary? 22 A. I have heard that, yes. 23 Q. To your knowledge would changes 24 from ELECT to COSTART be considered standard? Page 372 1 MR. MYERS: As part of this 2 standardization project? 3 MS. ZETTLER: Right, those types of 4 changes. 5 A. Again, I don't know if that was 6 part of that conversion. 7 Q. To your knowledge, has Prozac 8 ever been approved for a ten milligram dosage as 9 opposed to just a twenty milligram dosage? 10 A. Approved where? 11 Q. By the FDA. 12 A. By the FDA. For what 13 indication? 14 Q. It doesn't say here. It might -- 15 why don't we mark this then. 16 (PLAINTIFFS' EXHIBIT NO. 20 WAS 17 MARKED FOR IDENTIFICATION AND 18 RECEIVED IN EVIDENCE.) 19 A. To the best of my knowledge, 20 ten milligrams has been approved for depression. 21 Q. For depression, okay. Do you 22 recognize Exhibit 20? 23 A. Yes. 24 Q. What is it? Page 373 1 A. This is a news item that came 2 out over messenger E-Mail, it's part of what's 3 called Lilly News, LLY News. 4 Q. Is this document something that 5 is compiled at Lilly? 6 A. I believe it is. Someone at 7 Lilly tapes different articles and puts them 8 together. 9 Q. Different articles that they 10 believe people such as yourself would be 11 interested in from your employment standpoint? 12 A. That's correct. 13 Q. Do you know why Lilly wanted to 14 or sought to have a ten milligram dosage approved 15 by the FDA? 16 A. I'm not sure explicitly, no, 17 I'm not sure explicitly. It has to do with 18 medical use, and I'll defer to physicians on 19 that. 20 Q. Do you have any idea? 21 A. I sense that the physicians and 22 patients wanted ten milligrams. 23 Q. Does it have anything to do 24 with the side effect profile on Fluoxetine? Page 374 1 A. I'm not certain. 2 Q. Have you worked on any trials 3 or studies or analysis of data at Lilly that 4 compares a ten milligram dose of Prozac as 5 opposed to a twenty milligram or higher from the 6 standpoint of occurrence of adverse events? 7 A. I'm not familiar with any such 8 comparison comparing ten milligrams alone versus 9 twenty milligrams alone, I have not done that. 10 (PLAINTIFFS' EXHIBIT NO. 21 WAS 11 MARKED FOR IDENTIFICATION AND 12 RECEIVED IN EVIDENCE.) 13 Q. Do you recognize Exhibit 21? 14 A. Yes. 15 Q. Can you tell me what it is? 16 A. It's an outline of variables in 17 a data base, in the suicidality data base, to 18 identify potential risk factors for suicidal 19 acts, ideations and occurrences of three and four 20 for the Ham-D three score. 21 Q. Is that the algorithm you were 22 talking about earlier? 23 A. No, it's not. 24 Q. These variables that -- in Page 375 1 other words, outcome definitions, variable one, 2 occurrence of an act, is that something that's 3 used then across the board? In other words you 4 can use variable one and you'll always know that 5 it's occurrence of an act? 6 A. No, that was just specific for 7 this particular analysis. 8 Q. That's what I meant, within the 9 analysis, within this analysis. 10 A. Yes, each variable always means 11 the same thing for this particular analysis 12 within that data base. 13 Q. Is this analysis related to the -- 14 or is this document related to the analysis of 15 data for the BC article? 16 A. It's not related to that 17 analysis, but it's using the same data base. 18 Q. Do you know which analysis 19 they're talking about in this document? 20 A. This analysis was undertaken to 21 see if there's any factors at base line or before 22 patients are treated which lead to higher rates 23 of suicidal acts or ideation. 24 Q. Such acts like previous suicide Page 376 1 attempts, things of that nature? 2 A. One here that I noticed here is 3 where the patients are taking Benzodiazepines or 4 chloral hydrates went up, the patients taking 5 Benzodiazepines have higher incidence rates than 6 those that do not. 7 Q. So this is more of a historical 8 analysis? 9 A. This is a retrospective 10 analysis of the thirty sixty-five, the randomized 11 double-blind data base. 12 Q. Prior to treatment with 13 Fluoxetine? 14 A. No. 15 Q. You're confusing me again. 16 A. It looks at variables that are 17 collected prior to treatment with Fluoxetine or 18 placebo or tricyclic, and then looks at 19 occurrence or non-occurrence of a suicidal act or 20 ideation after treatment with one of those 21 ramdomized treatments and seeing if there's a 22 relationship between the two. 23 Q. Okay. Is there a final report 24 or an article that you know of that was written Page 377 1 based on the analysis done with these variables? 2 A. No. 3 Q. Why not? Is this related to 4 the article we were talking about earlier that 5 was your special project? 6 A. No, this was different. I did 7 some analyses of this data base and presented the 8 findings to the medical cohort, and basically at 9 that time it just gave to them obvious 10 indications of what they knew already and have 11 pursued since then. For example, if my memory 12 serves me correctly, and I don't remember the 13 details, that patients taking Benzodiazepine at 14 baseline tend to have higher acts than those that 15 do not. So there's no surprises in that and 16 haven't been taken any further from that. 17 Q. Did you write up a written 18 report with your findings to submit to the 19 medical department? 20 A. I don't believe so. I remember 21 typing up tables with the results, but as far as 22 a written report, I don't believe there was an 23 actual document summarizing that. 24 Q. What was the title that you Page 378 1 gave those various tables, is there a general 2 title that you gave to those tables? 3 A. I don't know if I even put a 4 title on the table, I'm not certain. 5 Q. Do you remember about 6 approximately when you did this project? 7 A. This was before the advisory 8 committee hearing. 9 Q. Okay. Was it in preparation 10 for the advisory committee hearing? 11 A. This was part of that process, 12 yes. 13 Q. Have you ever heard the term 14 Prozac verbatims before? 15 A. I believe so. 16 Q. What are Prozac verbatims? 17 A. I do not know. 18 (PLAINTIFFS' EXHIBIT NO. 22 WAS 19 MARKED FOR IDENTIFICATION AND 20 RECEIVED IN EVIDENCE.) 21 MS. ZETTLER: I don't have any other 22 questions. Thank you, Doctor. 23 MR. MYERS: Paul? 24 MS. ZETTLER: While Paul is getting his Page 379 1 act together, let me just state for the record 2 that I don't have any questions at this time 3 provided there was a ruling somewhere down the 4 line that we can ask Mr. Enas questions about the 5 documents that were produced on August 21st 6 throughout the hearing that was held on that date 7 regarding confidentiality. I'm going to reserve 8 the right to that extent to bring him back and 9 ask him any questions that I think are necessary 10 and relevant to the document. 11 MR. MYERS: Well, I've made our 12 position clear. Let's go off the record for a 13 second. 14 (DISCUSSION OFF THE RECORD.) 15 Q. (BY MS. ZETTLER) Does Exhibit 16 22 refresh your recollection as to what Prozac 17 verbatims are, Doctor? 18 A. Yes. 19 Q. Can you tell me what they are? 20 A. Basically it's just a summary 21 of answers to various questions. There are many 22 questions that have been posed, and this is a 23 summary of the analyses that speak to those 24 questions. Page 380 1 Q. When you say questions that 2 have been posed, are you talking about questions 3 posed by any particular entity such as the FDA? 4 A. No, in fact, questions that I 5 posed or others may have posed to me. It looks 6 like, if I remember correctly, this is something 7 I typed up. 8 Q. Okay. Can I look at that again 9 real quick? 10 A. Starting with the second page, 11 I didn't type the first page, that's not related 12 to the Prozac verbatim. 13 Q. Okay, okay. So even though 14 they run numerically by Pz number, they are not 15 related? 16 A. That's correct. 17 Q. So these questions that are 18 listed on Pz 2465 300 through -- 19 A. Three thirty. 20 Q. I'm sorry, 330 through 337 21 questions that you either had or other people 22 have had? 23 A. Through 333. 24 Q. Okay. Page 381 1 A. Were questions that I had or 2 others had, but I tried to summarize those for my 3 own sake to try to get a handle on all the 4 questions that have been around. 5 Q. What about the rest of that 6 exhibit, are those other pages related to the 7 verbatim? 8 A. They aren't typed, and they 9 aren't -- so they aren't related necessarily to 10 the verbatims. 11 Q. Can you tell by reviewing the 12 rest of the pages of Exhibit 22 if they are 13 related? 14 A. These are responses from others 15 in my staff to those questions, they aren't -- 16 they weren't given by me personally, but someone 17 else went out and addressed the question and then 18 summarized it here. 19 Q. Somebody else attempted to 20 answer the questions that you had listed? 21 A. That's correct. 22 Q. May I see that again real 23 quick? Did you answer any of these questions 24 yourself or are the answers to these questions Page 382 1 listed in the verbatims given to you by other 2 people? 3 A. Some questions I answered 4 myself and others were -- the answers were given 5 by analyses performed by other individuals. 6 Q. Can you tell just by looking at 7 the verbatim themselves and the answers what the -- 8 which questions you answered? 9 A. I can't determine exactly, I 10 can pick out questions and answers of which I was 11 more intimately involved in the analysis. 12 Q. Okay. Can you give an example? 13 A. For example, question seven is 14 the use of Benzos during therapy equally 15 distributed across treatment groups. I don't 16 remember actually typing those myself, but I was 17 overseeing or keeping tabs, keeping communication 18 with the statistician or analyst who was actually 19 computing those values. 20 Q. Okay. 21 A. And they would relate those 22 back to me. For example, question two, are there 23 demographic baseline characteristics associated 24 with increased risk of suicidality in general. A Page 383 1 preliminary overview, as part of this answer, was 2 one that was based on this analysis that I 3 previously referred to in another exhibit. 4 Q. The exhibit where we were 5 talking about the variables? 6 A. Yes. 7 Q. It was 21? 8 A. So that would be one where I 9 was more intimately involved with the analysis 10 myself. 11 MS. ZETTLER: All right. That's all I 12 have. Thank you. 13 * * * * * * * * * * 14 CROSS EXAMINATION 15 BY MR. CLEMENTI: 16 Q. Doctor, my name is Paul 17 Clementi and I represent a number of doctors, 18 including Doctor Allen Miller and Doctor Dickie 19 Kay. Am I correct in assuming that you don't 20 know or know of a Doctor Allen Miller or Doctor 21 Dickie Kay? 22 A. I don't remember those names. 23 Q. Okay. Could you give me your 24 definition just briefly, if you even have one, of Page 384 1 what exclusion criteria means? 2 A. My own personal understanding 3 of that term -- 4 Q. Right. 5 A. -- is that it's used in 6 protocols to prospectively exclude patients that 7 the investigator does not want included in that 8 particular study, for whatever medical reason 9 there might be. 10 Q. At which stage is an exclusion 11 criteria used, is it before or after counting the 12 patients or before or after the placebo wash-out, 13 where is that done? 14 MR. MYERS: When you say counting the 15 patients, counting the patients for what purpose? 16 MR. CLEMENTI: Exactly, that's what I'm 17 asking. 18 Q. I'm asking where is the 19 exclusion criteria used. Some studies on the top 20 say there were one thousand patients in this 21 study, that's what the report says. This many 22 were in at this stage, this many were in at this 23 stage, and this many were in at this stage and 24 the number keeps going down. Is the exclusion Page 385 1 criteria used before the first number or after 2 the first number if that makes any sense to you? 3 A. My understanding is that the 4 exclusion criteria is used up front. Before a 5 patient is enrolled, they have to meet the 6 inclusion/exclusion criteria before they're even 7 entered into the study. 8 Q. Now I'm going to ask a couple 9 of questions about percentages, and if you can't 10 answer these questions from your point of view, 11 please say so and I'll understand perfectly. 12 There are some numbers in this, and I'll try to 13 make it real simple so I can follow along. If a 14 study begins with one thousand subjects, and 15 there's a two week wash-out stage, okay, then the 16 second stage starts and it's a placebo versus a 17 test drug stage, that stage lasts eight weeks. 18 After two weeks of the second stage, only five 19 hundred patients return to continue with the 20 process. Are you with me so far? 21 A. Let me clarify. A thousand 22 patients are enrolled and how many are randomized 23 or entered in the double-blind or comparison 24 stage of the trial? Page 386 1 Q. One thousand. One thousand of 2 them, every single one of them, made it through 3 the wash-out. 4 A. Every single one goes through 5 the wash-out stage, so all one thousand are 6 randomized. 7 Q. Right. 8 A. And then five hundred show up 9 at visit -- at the second visit within that 10 comparison period? 11 Q. Let me take it a step at a 12 time. I'll first start with the most simple and 13 move it that way. First, there's a thousand 14 people start the two-week wash-out period, okay. 15 Before they're randomized, five hundred don't 16 show up anymore, before they ever get this second 17 drug or possibility of a second drug. So we're 18 down to five hundred people, okay. Of those 19 remaining five hundred patients, they're all 20 randomized, and they're all given the placebo or 21 a test drug. They follow through the rest of the 22 entire program. Of those remaining five hundred 23 patients that completed the study, five of them 24 had, let's call it, back pain, okay. When you're Page 387 1 preparing results of the study, what percentage 2 of the people would you say had back pain? 3 A. Which treatment group did those 4 patients, five patients, have back pain, and was 5 that back pain observed before the randomization 6 or after the randomization? 7 Q. After the randomization. 8 A. Okay. 9 Q. They started out with no back 10 pain, a thousand patients. After two weeks 11 placebo wash-out, before randomization, five 12 hundred people were gone, five hundred were left. 13 At the end, those five hundred people all made it 14 through, and of those five hundred people only 15 five had back pain, never had it before, no one 16 else had it. Could you give a percentage of what 17 people had back pain in that study? Did that 18 make sense? 19 A. Yes. Now in that study I can 20 give you a percentage. Generally the most 21 important comparison is how many had back pain of 22 one group versus the other. But overall, it 23 would be five patients had back pain out of the 24 five hundred that were randomized in that study. Page 388 1 Q. Okay. So that would be one 2 percent? 3 A. Correct. 4 Q. As opposed to point five 5 percent which includes the extra five hundred who 6 didn't make it into the randomization? 7 A. Yes. That group is irrelevant 8 because they never were treated during the active 9 portion of the study, so it's apples and oranges, 10 it's five out of five hundred who got the 11 treatment, either the placebo or the 12 experimental. 13 Q. Okay. 14 A. But again, the most important 15 comparison is to tell me how many got back pain 16 on placebo out of, say it's an equal 17 randomization, two fifty patients compared to how 18 many got back pain out of experimental, out of 19 the two fifty on that. 20 Q. That part I understand as well, 21 I'm just tying to get the numbers down. Now 22 we've got a different trial, okay. Again, the 23 first stage is a wash-out stage, you've got a 24 thousand patients or subjects. Then as the Page 389 1 second stage begins, where they're being 2 randomized, you have five hundred patients 3 starting there again. That five hundred dropped 4 out again, okay. After a week of taking -- after 5 a week of the second stage, they're taking the 6 drug, the drug is either the placebo or the test 7 drug, after a week one hundred patients get 8 stomach aches, they drop-out. The other four 9 hundred finish the entire test period. Now what 10 percentage in this case could you say had the 11 stomach aches of the study, I want to use that 12 term. Of the entire study, how many people had 13 or what percentage had stomach aches? 14 A. None of the four hundred who 15 completed the study had stomach aches, just those 16 one hundred? 17 Q. That's correct. 18 A. Then it would be of the five 19 hundred patients that were randomized to either 20 placebo or the other treatment arm, one hundred 21 patients out of that five hundred had a stomach 22 ache. 23 Q. And dropped out after that 24 first week? Page 390 1 A. They may have dropped out, but 2 the incidence would be one hundred had a stomach 3 ache out of five hundred. 4 Q. That's correct. So at the very 5 end of the study, what percentage would you say 6 had stomach aches or would you -- how would it be 7 reported at the end, what would you say about how 8 many people had stomach aches, et cetera, in the 9 study? 10 A. One hundred of the five hundred 11 patients who were allocated to either placebo or 12 treatment had stomach aches. 13 Q. Okay. 14 A. So twenty percent. 15 Q. Are the patients who drop out 16 because of the stomach ache, are they followed up 17 at all after they drop out because of the stomach 18 ache? 19 MR. MYERS: Wait a minute. Let me 20 object to the form here. You've given him a very 21 crude and rough hypothetical, and, you know, you 22 haven't given him enough facts in giving your 23 hypothetical to answer that. 24 MR. CLEMENTI: Okay. Page 391 1 MR. MYERS: What do you mean by 2 follow-up? 3 MR. CLEMENTI: All right. 4 MR. MYERS: I was with you for the 5 numbers, we were doing good up to that point. 6 Q. Okay. Are there certain 7 reasons people leave a study after they have been 8 randomized? The answer is yes? 9 A. Yes. 10 Q. Is one of those reasons they 11 just don't show up anymore? 12 A. Yes. 13 Q. Is another one of those reasons 14 that they become sick and don't want to continue 15 with the study? 16 A. Possibly. There's a number of 17 reasons why patients don't continue with the 18 study. 19 Q. If a patient leaves the study 20 because, let's say, they become sick as opposed 21 to they never show up again, does somebody try to 22 call them back and/or confer with them at all and 23 say, later on, after they said they dropped out, 24 how they are at that point. Page 392 1 A. I'm not sure what the process 2 is within the medical department to ascertain 3 that, but that is noted that the patient dropped 4 out due to an adverse event or became sick versus 5 just lost to follow-up, there's a clear 6 distinction between those two situations. 7 MS. ZETTLER: And lack of efficacy is 8 one of those distinctions, too. 9 THE WITNESS: That's correct. 10 MR. CLEMENTI: That's all I have. 11 MS. ZETTLER: No further questions. 12 MR. MYERS: No questions. 13 (THE WITNESS WAS EXCUSED.) Page 393 1 COMMONWEALTH OF KENTUCKY ) 2 : ss COUNTY OF JEFFERSON ) 3 4 I, MARY KATHLEEN NOLD, A NOTARY PUBLIC IN 5 AND FOR THE STATE OF KENTUCKY AT LARGE, DO HEREBY 6 CERTIFY THAT THE FOREGOING TESTIMONY OF 7 DR. GREGORY ENAS 8 WAS TAKEN BEFORE ME AT THE TIME AND PLACE AS 9 STATED IN THE CAPTION; THAT THE WITNESS WAS FIRST 10 DULY SWORN TO TELL THE TRUTH, THE WHOLE TRUTH, 11 AND NOTHING BUT THE TRUTH; THAT THE SAID 12 PROCEEDINGS WERE TAKEN DOWN BY ME IN STENOGRAPHIC 13 NOTES AND AFTERWARDS TRANSCRIBED UNDER MY 14 DIRECTION; THAT IT IS A TRUE, COMPLETE AND 15 CORRECT TRANSCRIPT OF THE SAID PROCEEDINGS SO 16 HAD; THAT THE APPEARANCES WERE AS STATED IN THE 17 CAPTION. 18 WITNESS MY SIGNATURE THIS THE 27TH DAY OF 19 SEPTEMBER, 1993. 20 MY COMMISSION EXPIRES MARCH 10, 1994. 21 22 23 _________________________ MARY KATHLEEN NOLD 24 COURT REPORTER AND NOTARY PUBLIC STATE OF KENTUCKY AT LARGE Page 394 1 2 E R R A T A S H E E T 3 4 COMMONWEALTH OF KENTUCKY ) : SS 5 COUNTY OF JEFFERSON ) 6 7 8 I, GREGORY ENAS, PH.D, THE UNDERSIGNED 9 DEPONENT, HAVE THIS DATE READ THE FOREGOING PAGES 10 OF MY DEPOSITION AND WITH THE CHANGES NOTED 11 BELOW, IF ANY, THESE PAGES CONSTITUTE A TRUE AND 12 ACCURATE TRANSCRIPTION OF MY DEPOSITION GIVEN ON 13 THE SEPTEMBER 16 AND 17, 1993 AT THE TIME AND 14 PLACE STATED THEREIN. 15 PAGE NO. LINE NO. CHANGE REASON 16 PAGE NO. LINE NO. CHANGE REASON 17 18 19 20 21 22 23 _____________________________ 24 GREGORY ENAS, PH.D Page 395 1 SWORN TO AND SUBSCRIBED BEFORE ME THIS 2 _____ DAY OF __________, 1993. 3 _____________________________ NOTARY PUBLIC, STATE OF 4 KENTUCKY AT LARGE Page 396 1 2 3 4 5 6 7 8 9 10 Page 397 1 DIRECT............................................14 2 CROSS EXAMINATION BY MS. ZETTLER: ................125 3 CROSS EXAMINATION BY MR. CLEMENTI:................384 4 (PLAINTIFFS' EXHIBIT NO. 1.......................129 5 (PLAINTIFFS' EXHIBIT NO. 2 ......................139 6 PLAINTIFFS' EXHIBIT NO. 3........................173 7 PLAINTIFFS' EXHIBIT NO. 4........................195 8 PLAINTIFFS' EXHIBIT NO. 5........................200 9 PLAINTIFFS' EXHIBIT NO. 6........................242 10 PLAINTIFFS' EXHIBIT NO. 7........................266 11 PLAINTIFFS' EXHIBIT NO. 8........................268 12 PLAINTIFFS' EXHIBIT NO. 9........................272 13 PLAINTIFFS' EXHIBIT NO. 10.......................275 14 PLAINTIFFS' EXHIBIT NO. 11.......................286 15 PLAINTIFFS' EXHIBIT NO. 12.......................294 16 PLAINTIFFS' EXHIBIT NO. 13.......................297 17 PLAINTIFFS' EXHIBIT NO. 14.......................301 18 PLAINTIFFS' EXHIBIT NO. 15.......................306 19 PLAINTIFFS' EXHIBIT NO. 16.......................317 20 PLAINTIFFS' EXHIBIT NO. 17.......................340 21 PLAINTIFFS' EXHIBIT NO. 18.......................341 22 PLAINTIFFS' EXHIBIT NO. 19.......................344 23 PLAINTIFFS' EXHIBIT NO. 20.......................373 24 PLAINTIFFS' EXHIBIT NO. 21.......................375 Page 398 1 PLAINTIFFS' EXHIBIT NO. 22.......................379 2 COMMONWEALTH.....................................394 3 E R..............................................395 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Page 399