AN HISTORICAL VIEW OF THE MEDICAL-SOCIAL ASPECTS OF THE UGDP HARVEY C. KNOWLES, JR., M.D. CINCINNATI
Few medical studies have stirred up as much controversy as the University Group Diabetes Program, known as the UGDP. It was designed as a multicenter study in 1961 to determine the effects of treatment on prevention of complications in the middle-aged stable-type diabetic.1 Having been concerned with the study since its beginning, I shall try to give an historical view of the highlights of the controversy and end with fears I have on acceptance of some of the long term goals of the study. My remarks will be limited for the most part to the controversy surrounding unexpected findings in patients treated with tolbutamide.2 In the winter of 1958-59, members of a study section at the National Institutes of Arthritis and Metabolic Diseases (NIAMD) discussed the question of the influence of diabetic control on the course of vascular disease and raised the point as to whether a clinical trial might provide an answer. In addition, there was a discussion of tolbutamide, a blood sugar lowering drug, which had just become available. A small group of clinical investigators interested in diabetes met in the spring of 1959 in Atlantic City to discuss the feasability of mounting such a study in prospective form. Interest was sufficiently great to influence the Council of the NIAMD to provide travel funds for one to two years to allow the group to discuss the matter further. The purpose was to decide if such a study should be undertaken, and if so, to develop plans for it. After discussion for over a year a decision to go forward was made, and a grant application was prepared and submitted to the NIAMD for peer review. In passing, I should like to compare the size of the application at that time with what is required for one today (Fig. 1). In fact, in the original application, little attention was paid to data storage or analysis. Nevertheless, the Council action was favorable, and grants for seven years were given to seven institutions. Later, six more institutions were added to include a Coordinating Center for data storage and analytic activities. In particular, the initial purposes of the study were three. The first was to answer the question as to whether tolbutamide had a beneficial effect on vascular disease, the second to answer the question as to whether lowering blood sugar had a similar effect, and the third to 150
151
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
develop methods of clinical trials in diabetes. Tolbutamide was of great interest at the time, for it caused secretion of human insulin which passed through the liver prior to acting at the periphery. Exogenous insulin is derived from other species, and it can act without passing first through the liver. The design was to randomize five treatments among groups of 200 patients, each drawn from a population of middle-aged stable-type diabetics with diagnosis made within the past year. All received diet prescription, and the five treatment groups were tolbutamide, insulin for control, insulin for non-carbohydrate effect with use of a low dose, phenformin and placebo. Three of the treatments, tolbutamide, insulin for control, and diet alone as represented by placebo were in general use in diabetic practice at the time, and the patients entering the study could have received any of the three. Symptomatic patients were not included. Measurements of vascular and other response variables were standardized and made at designated intervals over time. Independent variables measured simultaneously were blood sugar and lipid levels, and body weight. Treatments were randomized within centers with the assumption of homogenous distributions of baseline risk factors between groups.
The study got under way in 1961. In 1968 a disturbing event was noted in that the cumulative rate of cardiovascular deaths in the tolbutamide group exceeded those in the insulin and placebo groups. Total deaths CUMULATIVE MORTALITY RATES PER 100 POPULATION AT RISK BY YEAR OF FOLLOW-UP 20
'eT0LIM
Cu,sn
Port A. All
20
Part a Covasular Cas
WZ~~~~~~~~~~~~~~~~2 PlSt.. ,;;evcto w
4
15~~~~~~~~~~~I IVAR
lo-
_j
YEARS OF FOLLOW-UP
lo v pL5 P180
sTOLD
eo
1~~~~~~~~~~~~~~~~~~~~~~1 I
,
IVA B "-L :1 ISTD~~~~~~~~~~~~~IT -~ ~ ~~~~~~~~~~~PB
YEARS Of FOLLOW-UP
FIG. 1. Cumulative Mortality rates per 100 population at risk by year of follow-up.
152
HARVEY C. KNOWLES, JR.
were increased also, but not at a significant level (Fig. 2). In 1969 the cardiovascular mortality increase was accepted as manifest by the investigators, and the drug was discontinued as a clinical treatment. It was the concensus that if harmful effects existed, and other treatment means were available, it would not be ethical to continue the drug under experimental circumstances. In the winter of 1970, the decision was made to submit the findings in abstract form to the American Diabetes Association for consideration of presentation at the June annual meeting in St. Louis. The abstract was forwarded in late February, and the Program Committee of the Association meeting in March in Chicago accepted it for presentation. Little did the investigators know what was in the offing. An unexpected chain of events was about to start. On May 20 at 2:17 PM, it came over the Dow-Jones wires that a drug used to lower blood sugar in diabetes might be harmful. The next day the Washington Post contained an article to the effect that 8,000 people might die in a year because of taking tolbutamide. This was not a particularly good announcement for the study, especially when appearing initially in the lay press. The same day the Wall Street Journal published an article on the study, but of a more academic nature. Both papers stated that pharmaceutical houses were challenging the study. It is not known how the press obtained acess to the study report. It was around the time, however, that the program of the annual meeting of the American Diabetes Association had been printed and mailed, which would have put the abstract in the public domain.
FIG. 2. Paperwork!
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
153
The tolbutamide findings were presented at the meeting in St. Louis and were followed by five speakers giving data indicating findings to the contrary. The meeting was indeed well attended, not only by physicians, but by stockbrokers and persons of other sorts as well. Discussion after the presentations was heated, and opposition to the findings predominated. The opposition has continued to this day. The UGDP findings were written up and submitted to DIABETES for publication. After extensive biometrical review the paper was published in the fall of 1970.2 An ad hoc committee appointed by the American Diabetes Association approved of the study at the time also (2). Nevertheless, the general opposition to the findings continued and has been acted out along several lines. Over the years these have included the following approaches: 1) Pharmaceutical Salesmen -Sales directions have included talks to medical students, house staff, and practicing physicians on the good effects of sulfonylureas, with added caution on interpretation of the UGDP findings. This is understandable since the approach is a customary one in American business. 2) Speakers at Graduate Programs -Many physicians not accepting the UGDP findings have taken part in graduate teaching programs. Speakers favoring the study are few, and speaker support has not always been available for them. This is also understandable. 3) General Physician Opinion -There has developed a general feeling that the findings on tolbutamide are invalid. This has been advanced by the views of prominent clinicians in diabetes and the graduate programs listed above. In deference to the study, however, it should be stated that few physicians have examined the report in depth, particularly the deliberations. In fact, many have arrived at opinions without ever having seen the report. This last item is understandable in view of the length and complexity of the report, and those unskilled in clinical trials have depended on the criticisms of others. 4) Private press -Privately printed medical publications have, in articles, interviews, editorials and letters to the editor, provided overwhelmingly an anti-UGDP atmosphere. Materials gathered loosely by this author in the years 1970-1976 have been con or pro as follows: Private Publications, 1970-1976 No.
Medical Tribune .................. ............. Internal Medicine News ........... ............. Disease Management ........................... New York Times .................. ............. Wall Street Journal .............. .............. Washington Post ............................... Boston Globe ..................................
33+ 29 4 7 9+2 33 3 4 1 11 1 1 1 52+
Con
Pro
42
10
0
1
154
HARVEY C. KNOWLES, JR.
Comments have ranged from quotations of doubt of measurements and data analysis, to claims if therapeutic nihilism, intent to mislead physicians, alteration of data, conduct of inferior medical practice, etc. 5) Medical Press - Publications in this media have also been critical of the UGDP, but not to the extent of the private press. The findings on tolbutamide were not accepted as conclusive in England and Canada. The JAMA has printed material both against3 and for4 the study, but of all the medical journals in this country, has had the most material favorable to it. Medical Publications, 1970-1976 No.
JAMA ......................................... .. Diabetes ......................................... ............... Clin. Pharm. & Therapy ........... Ann. Intern. Med ................................ Am. J. Med ....................................... CMAJ .......................................... Lancet ........................................... Brit. Med. J .....................................
Pro
1911 12 22
Con
8 1 1
1 1 1 1
1 1 1
1
6) Committee on Care -A large group of leading diabeticians have formed a committee to work toward the prevention of the Federal Drug Administration (FDA) action in adding a label to tolbutamide to the effect that the drug might have cardiovascular risks. 7) Legal Intervention - It has been proposed by some that the original data and even patient records be obtained for study to determine the validity and allow for reanalysis. Petitions in this line against the investigation were filed in 1975. If the courts were to act favorably on the petitions, there could be future problems in confidentiality in all clinical investigation. Individuals, organizations or institutions might have access to raw data, patient records, etc., of any clinical investigation. In 1974, the National Institutes of Health requested the Biometric Society to investigate and evaluate the tolbutamide findings. The Society appointed its own committee which included members from Western Europe as well. Evaluation was made also of studies giving findings opposite to that of the UGDP. The Committee spent two years in analyzing the data. Site visits in depth were made to the Coordinating Center at the University of Maryland and the Clinical Centers of Cincinnati and Boston. They reported that they could not find any factor other than tolbutamide to account for the increased mortality.5 Nevertheless, the private press remained skeptical of the investigation. In 1976, the National Institutes of Arthritis, Metabolism and Digestive Diseases requested the FDA to evaluate the study. This investigation is now in progress, but the private press is "wary" of the audit.
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
155
In 1976, DIABETES and THE NEW ENGLAND JOURNAL OF MEDICINE published identical editorials6 7 stating that sufficient evidence was now at hand indicating that closer control of diabetes would decrease the frequency of microvascular disease. Subsequently, only five letters addressed to the editor appeared, and no counter papers have been printed. Contrast this response with that to the findings with tolbutamide, the issue on control being of a far deeper nature in treatment of diabetes than that concerning one drug only. In 1964, I spoke to this Association about the need for new approaches to the treatment of diabetes. In 1976, I make the same plea. I believe that the UGDP has done much to redirect the attention to treatment of diabetics in the area of middle-aged stable-type diabetes wherein the large number of patients with coronary artery disease make diabetes a major health problem. Those working in the UGDP are concerned about future publications, however. The second purpose of the study was to determine the effect of the control of blood sugar on vascular disease. In addition to control, measurements of lipids, weight, and blood pressure over time have been made and now are being studied in relation to vascular response variables. These data and analyses could be of greater value to both investigators and clinicians. But the impact of the tolbutamide findings have caused such adversity that one may wonder if the reports in preparation on the general course of events over time and interactions of independent variables will be accepted. If the same skepticisms are shown, then money may have been spent and over ten years of work gone for naught. Hopefully, the data will be acceptable. REFERENCES 1. Prepared by KLIMT, CR, KNATTERUD, GL, MEINERT, CL, PROUT, TE: The University Group Diabetes Program: A Study of the Effects of Hypoglycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. Part I: Design, Methods and
Baseline Characteristics. Diabetes 19: Suppl. 2, 1970. 2. Prepared by KLIMT, CR, KNATTERUD, GL, MEINERT, CL, PROUT, TE: The University Group Diabetes Program: A Study of the Effects of Hypoglycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. Part II: Mortality Results.
Diabetes 19: Suppl. 2, 1970. 3. SCHOR, S: The University Group Diabetes Program. A Statistician Looks at the Mortality Results. JAMA 217: 1671, 1971. 4. CORNFIELD, J: The University Group Diabetes Program. A Further Statistical Analysis of the Mortality Findings. JAMA 217: 1676, 1971. 5. Report of the Committee for the Assessment of Biometric Aspects of Controlled Trials of Hypoglycemic Agents. JAMA 231: 583, 1975. 6. CAHILL, GF, JR, ETZWILER, DD, FREINKEL, N: Blood Glucose Control in Diabetes. Diabetes 25: 237, 1976. 7. CAHILL, GF, JR: 'Control' and Diabetes. New Eng J Med 294: 1004, 1976.
DISCUSSION DR. HERBERT G. LANGFORD (Jackson): You didn't emphasize, but I think one of the critical causes of the immediate response was this unfortunate leak, some six weeks
156
HARVEY C. KNOWLES, JR.
before the diabetes meetings. As I talked to some of the biggest opponents, I find their opinion was fixed that this was phoney before they could even get hold of the abstract, and I do think you have to examine the data very carefully. I went through the whole doggone thing, and it's pretty good. In fact, it's amazingly good. Actually the report proves that randomization works, and you don't have to stratify on all the variables. But, for future trials, we have to be sure that everything is absolutely locked up until we can publish all the facts. DR. KNOWLEs: Thank you Dr. Langford, for those kind words, we have spoken very openly about the problem of the leak for years. I can only say that the deadline for abstracts was March 1. In February we met in the Washington Air Port and decided to send an abstract to the 1970 St. Louis meeting. It was received in New York on March 1. The program was mailed somewhere in the middle of May. We are unable to establish this exact date. But it could be very well that the day after they went in the mail, that they were picked up, because it was May 21, when the first news came over. It must have come from programs.
Few medical studies have stirred up as much controversy as the University Group Diabetes Program, known as the UGDP. It was designed as a multicenter study in 1961 to determine the effects of treatment on prevention of complications in the middle-aged stable-type diabetic.1 Having been concerned with the study since its beginning, I shall try to give an historical view of the highlights of the controversy and end with fears I have on acceptance of some of the long term goals of the study. My remarks will be limited for the most part to the controversy surrounding unexpected findings in patients treated with tolbutamide.2 In the winter of 1958-59, members of a study section at the National Institutes of Arthritis and Metabolic Diseases (NIAMD) discussed the question of the influence of diabetic control on the course of vascular disease and raised the point as to whether a clinical trial might provide an answer. In addition, there was a discussion of tolbutamide, a blood sugar lowering drug, which had just become available. A small group of clinical investigators interested in diabetes met in the spring of 1959 in Atlantic City to discuss the feasability of mounting such a study in prospective form. Interest was sufficiently great to influence the Council of the NIAMD to provide travel funds for one to two years to allow the group to discuss the matter further. The purpose was to decide if such a study should be undertaken, and if so, to develop plans for it. After discussion for over a year a decision to go forward was made, and a grant application was prepared and submitted to the NIAMD for peer review. In passing, I should like to compare the size of the application at that time with what is required for one today (Fig. 1). In fact, in the original application, little attention was paid to data storage or analysis. Nevertheless, the Council action was favorable, and grants for seven years were given to seven institutions. Later, six more institutions were added to include a Coordinating Center for data storage and analytic activities. In particular, the initial purposes of the study were three. The first was to answer the question as to whether tolbutamide had a beneficial effect on vascular disease, the second to answer the question as to whether lowering blood sugar had a similar effect, and the third to 150
151
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
develop methods of clinical trials in diabetes. Tolbutamide was of great interest at the time, for it caused secretion of human insulin which passed through the liver prior to acting at the periphery. Exogenous insulin is derived from other species, and it can act without passing first through the liver. The design was to randomize five treatments among groups of 200 patients, each drawn from a population of middle-aged stable-type diabetics with diagnosis made within the past year. All received diet prescription, and the five treatment groups were tolbutamide, insulin for control, insulin for non-carbohydrate effect with use of a low dose, phenformin and placebo. Three of the treatments, tolbutamide, insulin for control, and diet alone as represented by placebo were in general use in diabetic practice at the time, and the patients entering the study could have received any of the three. Symptomatic patients were not included. Measurements of vascular and other response variables were standardized and made at designated intervals over time. Independent variables measured simultaneously were blood sugar and lipid levels, and body weight. Treatments were randomized within centers with the assumption of homogenous distributions of baseline risk factors between groups.
The study got under way in 1961. In 1968 a disturbing event was noted in that the cumulative rate of cardiovascular deaths in the tolbutamide group exceeded those in the insulin and placebo groups. Total deaths CUMULATIVE MORTALITY RATES PER 100 POPULATION AT RISK BY YEAR OF FOLLOW-UP 20
'eT0LIM
Cu,sn
Port A. All
20
Part a Covasular Cas
WZ~~~~~~~~~~~~~~~~2 PlSt.. ,;;evcto w
4
15~~~~~~~~~~~I IVAR
lo-
_j
YEARS OF FOLLOW-UP
lo v pL5 P180
sTOLD
eo
1~~~~~~~~~~~~~~~~~~~~~~1 I
,
IVA B "-L :1 ISTD~~~~~~~~~~~~~IT -~ ~ ~~~~~~~~~~~PB
YEARS Of FOLLOW-UP
FIG. 1. Cumulative Mortality rates per 100 population at risk by year of follow-up.
152
HARVEY C. KNOWLES, JR.
were increased also, but not at a significant level (Fig. 2). In 1969 the cardiovascular mortality increase was accepted as manifest by the investigators, and the drug was discontinued as a clinical treatment. It was the concensus that if harmful effects existed, and other treatment means were available, it would not be ethical to continue the drug under experimental circumstances. In the winter of 1970, the decision was made to submit the findings in abstract form to the American Diabetes Association for consideration of presentation at the June annual meeting in St. Louis. The abstract was forwarded in late February, and the Program Committee of the Association meeting in March in Chicago accepted it for presentation. Little did the investigators know what was in the offing. An unexpected chain of events was about to start. On May 20 at 2:17 PM, it came over the Dow-Jones wires that a drug used to lower blood sugar in diabetes might be harmful. The next day the Washington Post contained an article to the effect that 8,000 people might die in a year because of taking tolbutamide. This was not a particularly good announcement for the study, especially when appearing initially in the lay press. The same day the Wall Street Journal published an article on the study, but of a more academic nature. Both papers stated that pharmaceutical houses were challenging the study. It is not known how the press obtained acess to the study report. It was around the time, however, that the program of the annual meeting of the American Diabetes Association had been printed and mailed, which would have put the abstract in the public domain.
FIG. 2. Paperwork!
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
153
The tolbutamide findings were presented at the meeting in St. Louis and were followed by five speakers giving data indicating findings to the contrary. The meeting was indeed well attended, not only by physicians, but by stockbrokers and persons of other sorts as well. Discussion after the presentations was heated, and opposition to the findings predominated. The opposition has continued to this day. The UGDP findings were written up and submitted to DIABETES for publication. After extensive biometrical review the paper was published in the fall of 1970.2 An ad hoc committee appointed by the American Diabetes Association approved of the study at the time also (2). Nevertheless, the general opposition to the findings continued and has been acted out along several lines. Over the years these have included the following approaches: 1) Pharmaceutical Salesmen -Sales directions have included talks to medical students, house staff, and practicing physicians on the good effects of sulfonylureas, with added caution on interpretation of the UGDP findings. This is understandable since the approach is a customary one in American business. 2) Speakers at Graduate Programs -Many physicians not accepting the UGDP findings have taken part in graduate teaching programs. Speakers favoring the study are few, and speaker support has not always been available for them. This is also understandable. 3) General Physician Opinion -There has developed a general feeling that the findings on tolbutamide are invalid. This has been advanced by the views of prominent clinicians in diabetes and the graduate programs listed above. In deference to the study, however, it should be stated that few physicians have examined the report in depth, particularly the deliberations. In fact, many have arrived at opinions without ever having seen the report. This last item is understandable in view of the length and complexity of the report, and those unskilled in clinical trials have depended on the criticisms of others. 4) Private press -Privately printed medical publications have, in articles, interviews, editorials and letters to the editor, provided overwhelmingly an anti-UGDP atmosphere. Materials gathered loosely by this author in the years 1970-1976 have been con or pro as follows: Private Publications, 1970-1976 No.
Medical Tribune .................. ............. Internal Medicine News ........... ............. Disease Management ........................... New York Times .................. ............. Wall Street Journal .............. .............. Washington Post ............................... Boston Globe ..................................
33+ 29 4 7 9+2 33 3 4 1 11 1 1 1 52+
Con
Pro
42
10
0
1
154
HARVEY C. KNOWLES, JR.
Comments have ranged from quotations of doubt of measurements and data analysis, to claims if therapeutic nihilism, intent to mislead physicians, alteration of data, conduct of inferior medical practice, etc. 5) Medical Press - Publications in this media have also been critical of the UGDP, but not to the extent of the private press. The findings on tolbutamide were not accepted as conclusive in England and Canada. The JAMA has printed material both against3 and for4 the study, but of all the medical journals in this country, has had the most material favorable to it. Medical Publications, 1970-1976 No.
JAMA ......................................... .. Diabetes ......................................... ............... Clin. Pharm. & Therapy ........... Ann. Intern. Med ................................ Am. J. Med ....................................... CMAJ .......................................... Lancet ........................................... Brit. Med. J .....................................
Pro
1911 12 22
Con
8 1 1
1 1 1 1
1 1 1
1
6) Committee on Care -A large group of leading diabeticians have formed a committee to work toward the prevention of the Federal Drug Administration (FDA) action in adding a label to tolbutamide to the effect that the drug might have cardiovascular risks. 7) Legal Intervention - It has been proposed by some that the original data and even patient records be obtained for study to determine the validity and allow for reanalysis. Petitions in this line against the investigation were filed in 1975. If the courts were to act favorably on the petitions, there could be future problems in confidentiality in all clinical investigation. Individuals, organizations or institutions might have access to raw data, patient records, etc., of any clinical investigation. In 1974, the National Institutes of Health requested the Biometric Society to investigate and evaluate the tolbutamide findings. The Society appointed its own committee which included members from Western Europe as well. Evaluation was made also of studies giving findings opposite to that of the UGDP. The Committee spent two years in analyzing the data. Site visits in depth were made to the Coordinating Center at the University of Maryland and the Clinical Centers of Cincinnati and Boston. They reported that they could not find any factor other than tolbutamide to account for the increased mortality.5 Nevertheless, the private press remained skeptical of the investigation. In 1976, the National Institutes of Arthritis, Metabolism and Digestive Diseases requested the FDA to evaluate the study. This investigation is now in progress, but the private press is "wary" of the audit.
MEDICAL-SOCIAL ASPECTS OF UGDP: HISTORICAL VIEW
155
In 1976, DIABETES and THE NEW ENGLAND JOURNAL OF MEDICINE published identical editorials6 7 stating that sufficient evidence was now at hand indicating that closer control of diabetes would decrease the frequency of microvascular disease. Subsequently, only five letters addressed to the editor appeared, and no counter papers have been printed. Contrast this response with that to the findings with tolbutamide, the issue on control being of a far deeper nature in treatment of diabetes than that concerning one drug only. In 1964, I spoke to this Association about the need for new approaches to the treatment of diabetes. In 1976, I make the same plea. I believe that the UGDP has done much to redirect the attention to treatment of diabetics in the area of middle-aged stable-type diabetes wherein the large number of patients with coronary artery disease make diabetes a major health problem. Those working in the UGDP are concerned about future publications, however. The second purpose of the study was to determine the effect of the control of blood sugar on vascular disease. In addition to control, measurements of lipids, weight, and blood pressure over time have been made and now are being studied in relation to vascular response variables. These data and analyses could be of greater value to both investigators and clinicians. But the impact of the tolbutamide findings have caused such adversity that one may wonder if the reports in preparation on the general course of events over time and interactions of independent variables will be accepted. If the same skepticisms are shown, then money may have been spent and over ten years of work gone for naught. Hopefully, the data will be acceptable. REFERENCES 1. Prepared by KLIMT, CR, KNATTERUD, GL, MEINERT, CL, PROUT, TE: The University Group Diabetes Program: A Study of the Effects of Hypoglycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. Part I: Design, Methods and
Baseline Characteristics. Diabetes 19: Suppl. 2, 1970. 2. Prepared by KLIMT, CR, KNATTERUD, GL, MEINERT, CL, PROUT, TE: The University Group Diabetes Program: A Study of the Effects of Hypoglycemic Agents on Vascular Complications in Patients with Adult-Onset Diabetes. Part II: Mortality Results.
Diabetes 19: Suppl. 2, 1970. 3. SCHOR, S: The University Group Diabetes Program. A Statistician Looks at the Mortality Results. JAMA 217: 1671, 1971. 4. CORNFIELD, J: The University Group Diabetes Program. A Further Statistical Analysis of the Mortality Findings. JAMA 217: 1676, 1971. 5. Report of the Committee for the Assessment of Biometric Aspects of Controlled Trials of Hypoglycemic Agents. JAMA 231: 583, 1975. 6. CAHILL, GF, JR, ETZWILER, DD, FREINKEL, N: Blood Glucose Control in Diabetes. Diabetes 25: 237, 1976. 7. CAHILL, GF, JR: 'Control' and Diabetes. New Eng J Med 294: 1004, 1976.
DISCUSSION DR. HERBERT G. LANGFORD (Jackson): You didn't emphasize, but I think one of the critical causes of the immediate response was this unfortunate leak, some six weeks
156
HARVEY C. KNOWLES, JR.
before the diabetes meetings. As I talked to some of the biggest opponents, I find their opinion was fixed that this was phoney before they could even get hold of the abstract, and I do think you have to examine the data very carefully. I went through the whole doggone thing, and it's pretty good. In fact, it's amazingly good. Actually the report proves that randomization works, and you don't have to stratify on all the variables. But, for future trials, we have to be sure that everything is absolutely locked up until we can publish all the facts. DR. KNOWLEs: Thank you Dr. Langford, for those kind words, we have spoken very openly about the problem of the leak for years. I can only say that the deadline for abstracts was March 1. In February we met in the Washington Air Port and decided to send an abstract to the 1970 St. Louis meeting. It was received in New York on March 1. The program was mailed somewhere in the middle of May. We are unable to establish this exact date. But it could be very well that the day after they went in the mail, that they were picked up, because it was May 21, when the first news came over. It must have come from programs.
