- Email: [email protected]
ROLE OF THE PHARMACEUTICAL INDUSTRY IN MAJOR CLINICAL TRIALS
1258
the factors that discouraged referral of mothers to specialist
Clinical Trials
centres.
More staff and equipment for perinatal and neonatal care at King’s Mill Hospital might therefore have reduced the residual neonatal deaths in the community, although higher rates of handicap would have been expected among ventilated survivors who weighed less than 1000 g at births Epidemiological evidence, however, reveals that introduction of such measures does not necessarily reduce early neonatal mortality,2o,21 and pathological evidence shows that it does not reduce the proportion of neonatal deaths due to hyaline membrane disease or intraventricular haemorrhage.22 Thus clinical, statistical, pathological, and epidemiological observations all cast doubt on the value of sustained ventilation of the newborn as practised at present; only large randomised controlled trials will be able to determine whether its hazards outweigh its benefits to a community. Such trials are not impossible to mount but will no doubt be resisted. The recent history of perinatal medicine abounds with instances in which belated controlled trials eventually revealed that the apparent benefits of some widely acclaimed treatment had merely disguised the real extent of its tragic consequences.23 In the meantime clinicians await clearer guidance on how to identify those newborn infants who will benefit from sustained mechanical ventilation. I thank the medical and nursing staffs of the neonatal nurseries at Nottingham City Hospital, Derby City Hospital, and Scarsdale Hospital, Chesterfield, for access to residential and medical information on patients in their care. I also thank the many members of the medical and nursing staffs of the regional neonatal referral units at Nottingham City Hospital and the Jessop Hospital, Sheffield, for their devotion, skills, and efficiency. Necropsies were performed by Dr D. G. Fagan, the late Dr A. Macfarlane, Dr S. Variend, and Dr Joan D. Webb. REFERENCES 1. Perinatal and neonatal mortality. House of Commons second report from the Social Services Committee. 1977-80. Vol 1. London: HM Stationery Office, 1980: 35-39. 2. Walker CHM. Neonatal intensive care and stress. Arch Dis Child 1982; 57: 85-88 3. Phibbs CS, Williams RL, Phibbs RH. Newborn risk factors and costs of neonatal care. Pediatrics 1981; 68: 313-21. 4. Schroeder SE, Showstack JA, Roberts HE. Frequency and clinical description of
high-cost patients in
17 acute care hospitals.
N Engl J Med 1979; 300: 1306-09.
5. Haas R, Davies PA. Iatrogenic hazards in the newborn intensive care unit. In Wharton B, ed. Topics in perinatal medicine. Tunbridge Wells. Pitman Medical, 1980: 104-19. 6. Murdock AI, Linsao L, Reid MMcC, et al. Mechanical ventilation in the respiratory distress syndrome: A controlled trial. Arch Dis Child 1970; 45: 624-33. 7. Reid DHS, Tunstall ME, Mitchell RG. A controlled trial of artificial respiration in the respiratory-distress syndrome of the newborn. Lancet 1967; i: 532-33. 8. Silverman WA, Sinclair JC, Gandy GM, Finster M, Bauman WA, Agate FJ Jr. A controlled trial of management of respiratory distress syndrome m a body-enclosing respirator. I. Evaluation of safety. Pediatrics 1967; 39: 740-48. 9. Sinclair JC, Engel K, Silverman WA. Early correction of hypoxaemia and acidaemia in infants of low birth weight: A controlled trial of oxygen breathing, rapid alkali infusion and assisted ventilation. Pediatrics 1968; 42: 565-88. 10. Macfarlane A, Mugford M. Birth counts: Statistics of pregnancy and childbirth. Table A 7·1 Sizes of maternity units and numbers of births, England, 1973, 1978, 1980. National Perinatal Epidemiology Unit. London: HM Stationery Office, 1984: 197. 11. Nottinghamshire Area Health Authority (T) (Planning Dept). 12. Clayton SG, Lewis TLT, Pinker G, eds. Obstetrics by ten teachers. London: Edward Arnold, 1980: 114-20. 13. Roberton NRC. A manual of neonatal intensive care London. Edward Arnold, 1981: 35-44. 14. Steiner ES, Sanders EM, Phillips ECK. Maddock CR. Very low birth weight children at school age: Comparison of neonatal management methods. Br Med J 1980; 281: 1237-40. 15. Vyas H, Milner AD. Other respiratory diseases in the neonate. In: Roberton NRC ed. Textbook of neonatology Edinburgh: Churchill Livingstone, 1986: 312-39. 16. Roberton NRC. The care of neonates with respiratory failure. In: Chiswick ML ed. Recent advances in perinatal medicine I. London: Churchill Livingstone, 1983: 179. 17. Jones RM, Rutter N, Cooper AC, Pullan CR Pneumothorax in the neonatal period. Anaesthesia 1983; 38: 948-52 18. Cooke RWI. In utero transfer to specialist centres Arch Dis Child 1983; 58: 483 -84. 19. Levene MI, Dubowitz LMS. Long term follow-up of low birth weight babies. Br J Hosp Med 1982, 28: 487-93
ROLE OF THE PHARMACEUTICAL INDUSTRY IN MAJOR CLINICAL TRIALS D. G. JULIAN J. R. HAMPTON Department of Medicine, University Hospital, Queen’s Medical Centre, Nottingham NG7 2UH; and Department of Cardiology,
University of Newcastle upon Tyne Mos’r large clinical trials
are financially supported by the pharmaceutical industry. Grant-giving bodies have limited funds and might well feel that projects other than expensive clinical trials will give a better academic return, and large trials are expensive. The industry perceives the financial
risk to be worthwhile because the returns from a successful trial can be high. Relations between investigators and the sponsoring company are usually harmonious but this is not always so, and a code of practice might prove helpful. The results of a trial must be credible if they are to be widely accepted and form a basis for clinical practice. In general the less the industry is seen to be involved, the more credible the trial will be. However, the drug company will be paying a large sum of money: to demonstrate a reduction in mortality from cardiovascular disease a budget of$1 million would be modest,$3 or$4 million is probably typical, and the most expensive of such trials has cost over$100 million. A company investing on this scale will be hesitant about handing over total control of a trial to the investigators. Furthermore the company may fear that’dull" data, needed for submission to regulatory authorities, may not be properly collected or analysed by the investigators. The problem is to balance the company’s desire to obtain what it has paid for against the investigators need for independence from
commercial pressure.
During the design phase the sponsoring company needs to be closely involved so that it can be assured that the trial will answer the question it is interested in. At this stage either the investigators the company can withdraw with no more than time wasted and feathers ruffled. The trial protocol must define in detail how the company will be involved in the running and analysis of the study, because it may be too late to prevent misunderstandings once the trial has begun. The company, for example, must retain the right to terminate the trial if it learns from other studies that the drug has undesirable side-effects, or if it becomes apparent that too few endpoints are occurring for the trial to be brought to a conclusion within an acceptable time. or
A crucial design decision concerns the place to which data will sent for recording on computer. The trial’s treatment code is most easily added when the information is computerised, so whoever has access to the data-recording system will (at least in theory) have access to the results as the trial proceeds. Investigators are well aware of the statistical dangers of "multiple looks" at
be
20. 21.
Eksmyr R, Eklund G. Early neonatal deaths in geographically defined populations with different organisation of medical care Acta Paediatr Scand 1985; 74: 848-54. Eksmyr R. Two geographically defined populations with different organisation of medical care. Comparison of perinatal risks. Acta Paediatr Scand 1985; 74:
855-60. 22. Barson AJ, Tasker M, Lieberman BA, Hiller VF. Impact of improved perinatal care on the causes of death. Arch Dis Child 1984; 59: 199-207. 23. Silverman WA. Retrolental fibroplasia: A modern parable. New York. Grune and Stratton, 1980: 28, 37-42, 69-89.
1259 the results of a trial in progress, but if this information is available to a company it might attempt to terminate the trial early, either because an apparently’good" result had appeared, which would be commercially beneficial, or because an apparently‘ ‘bad" trend suggested that money was being wasted.
Although most pharmaceutical companies behave properly in trials and towards investigators, there have been isolated examples of unethical conduct. For example, companies have tried to prevent publication of unfavourable results by pressure on investigators or on journals. One sponsoring company with access to data on a trial in progress found that the drug was proving ineffective, and so terminated the study; then the company used data from uncontrolled trials for advertising purposes. Another company used the results from a single centre in a multicentre trial to support its application for a drug licence, even though the results from the other centres were unfavourable. Most seasoned investigators have a personal blacklist of companies with whom they feel they cannot do business.
Investigators’ motives and actions are not always pure either: are plenty of examples of both academic and financial skulduggery perpetrated by trialists. The sponsor, like the investigators, has a right to expect honesty. It is thus important, both from the investigators’ and the company’spoint of view, that
there
the trial results should be held on a computer to which neither investigators nor company has access. It is also important to define in advance criteria for stopping a trial earlier than initially planned.
If the trial is to be seen to be immune to"multiple looks" by the investigators and free from interference by the sponsoring company, the data must be handled by an independent data centre and the results must be periodically reviewed by an ethical committee that is wholly independent of both the company, the investigators, and (so far as is
possible) the data centre itself. The data centre should probably be an academic unit with a reputation to maintain. It seems unavoidable that the company, even though it possesses adequate computing facilities, will have to fund the data
centre. The centre must agree with the company that data will be recorded in a way that is acceptable to the sponsor and which is compatible with the company’s own computer, so that at the end of the trial the sponsor can make any analysis it wishes. There may sometimes be room for compromise: for example, some of the routine data which are not directly important for the trial could be separated and retained in the company’s computer without the treatment code.
If the data handling problems are so large that the company’s computer must be used, then data should be available to the data centre for the intermediate analyses required by the ethical committee. The ethical committee must be independent, and it will only be seen to be so if it receives information from the data centre and not from the company. The ethical committee is the guardian of the trial. It protects the patients by ensuring that they are not exposed to harm nor deprived of benefit-ie, it applies the "stopping rules" if the results become unequivocal before the forecasted end. It protects the investigators by relieving them of the responsibility of unwittingly harming their patients or depriving them of benefit. And it also protects the sponsor by insisting on up-to-date information and by having the right to inspect the data centre if it has any reason to believe that data has not been properly collected. The ethical committee, whose composition must be agreed beforehand by the investigators and the sponsor, must insist on its independence from both parties once the trial has
begun.
Drug regulatory authorities usually hold the pharmaceutical company responsible for reporting adverse reactions to drugs (ADR). A company may therefore claim with some justification that it must hold the treatment code and receive all ADR reports. If an ADR is unexpected this is not important, but if, for example, we include the events expected in a cardiovafcular trial (such as death, myocardial infarction, and stroke) then it is clearly undesirable for the company to be privy to this information.
Ideally the ADRs who can supply to the company any "unexpected" ADRs associated with active treatment so that these can be added to the database the company will have built up from other studies. Such ADRs should be notified to the company quickly. "Expected" ADRs should, however, only be notified to the ethical committee, who should be responsible for informing the licensing authorities if any significant imbalance develops between treatment groups. This implies a delegation of responsibility from the licensing authority to the ethical committee; this has been permitted in some trials in the UK but is not official government policy. should be
sent to
the data
centre
Although the sponsor should not have access to trial data, it has an important role in ensuring that the study is properly run and the data properly collected. Non-medical trial monitors, employed by the company, will ensure continuity of drug supplies and any necessary technical facilities, and should visit the participating centres regularly to identify problems at an early stage. Because these monitors will have no access to the data, they will not be able to exert undesirable pressures; relations with the investigators can and should be close. The role of the monitors is especially important in international multicentre trials, because the international co-ordinator is unlikely to be familiar with foreign medical practices and will probably not know many of the investigators. However, even then the monitor must not be responsible for the trial conduct in his country; this must be a job for one of the investigators who will take on the role of national co-ordinator. Data must be sent through the national co-ordinator to the international co-ordinator and hence to the data centre, so that the co-ordinators at both levels can ensure a proper standard of recording. The final analysis must be done by the investigators independently of the sponsor. Although the company should have the right to comment on any manuscript, the ultimate decision on publication must be for the investigators and a journal editor. The results of all studies must be published, however briefly, because "pooling" (meta-analysis) is worthless if only positive trials are considered. To ensure that results are eventually available a register of trials could be established either with the regulatory authority or with a journal, and the publication of abstracts of protocols of major trials would alert investigators to seek the results if a publicaton did not appear. A journal editor will decide whether the design and conduct of a trial has been of a sufficiently high standard to merit publication. An editor will only feel confident that both the investigators and the sponsoring company have behaved with propriety if the appropriate organisation was built into the trial during its design. Correspondence should be addressed
to
K. R. H.
the factors that discouraged referral of mothers to specialist
Clinical Trials
centres.
More staff and equipment for perinatal and neonatal care at King’s Mill Hospital might therefore have reduced the residual neonatal deaths in the community, although higher rates of handicap would have been expected among ventilated survivors who weighed less than 1000 g at births Epidemiological evidence, however, reveals that introduction of such measures does not necessarily reduce early neonatal mortality,2o,21 and pathological evidence shows that it does not reduce the proportion of neonatal deaths due to hyaline membrane disease or intraventricular haemorrhage.22 Thus clinical, statistical, pathological, and epidemiological observations all cast doubt on the value of sustained ventilation of the newborn as practised at present; only large randomised controlled trials will be able to determine whether its hazards outweigh its benefits to a community. Such trials are not impossible to mount but will no doubt be resisted. The recent history of perinatal medicine abounds with instances in which belated controlled trials eventually revealed that the apparent benefits of some widely acclaimed treatment had merely disguised the real extent of its tragic consequences.23 In the meantime clinicians await clearer guidance on how to identify those newborn infants who will benefit from sustained mechanical ventilation. I thank the medical and nursing staffs of the neonatal nurseries at Nottingham City Hospital, Derby City Hospital, and Scarsdale Hospital, Chesterfield, for access to residential and medical information on patients in their care. I also thank the many members of the medical and nursing staffs of the regional neonatal referral units at Nottingham City Hospital and the Jessop Hospital, Sheffield, for their devotion, skills, and efficiency. Necropsies were performed by Dr D. G. Fagan, the late Dr A. Macfarlane, Dr S. Variend, and Dr Joan D. Webb. REFERENCES 1. Perinatal and neonatal mortality. House of Commons second report from the Social Services Committee. 1977-80. Vol 1. London: HM Stationery Office, 1980: 35-39. 2. Walker CHM. Neonatal intensive care and stress. Arch Dis Child 1982; 57: 85-88 3. Phibbs CS, Williams RL, Phibbs RH. Newborn risk factors and costs of neonatal care. Pediatrics 1981; 68: 313-21. 4. Schroeder SE, Showstack JA, Roberts HE. Frequency and clinical description of
high-cost patients in
17 acute care hospitals.
N Engl J Med 1979; 300: 1306-09.
5. Haas R, Davies PA. Iatrogenic hazards in the newborn intensive care unit. In Wharton B, ed. Topics in perinatal medicine. Tunbridge Wells. Pitman Medical, 1980: 104-19. 6. Murdock AI, Linsao L, Reid MMcC, et al. Mechanical ventilation in the respiratory distress syndrome: A controlled trial. Arch Dis Child 1970; 45: 624-33. 7. Reid DHS, Tunstall ME, Mitchell RG. A controlled trial of artificial respiration in the respiratory-distress syndrome of the newborn. Lancet 1967; i: 532-33. 8. Silverman WA, Sinclair JC, Gandy GM, Finster M, Bauman WA, Agate FJ Jr. A controlled trial of management of respiratory distress syndrome m a body-enclosing respirator. I. Evaluation of safety. Pediatrics 1967; 39: 740-48. 9. Sinclair JC, Engel K, Silverman WA. Early correction of hypoxaemia and acidaemia in infants of low birth weight: A controlled trial of oxygen breathing, rapid alkali infusion and assisted ventilation. Pediatrics 1968; 42: 565-88. 10. Macfarlane A, Mugford M. Birth counts: Statistics of pregnancy and childbirth. Table A 7·1 Sizes of maternity units and numbers of births, England, 1973, 1978, 1980. National Perinatal Epidemiology Unit. London: HM Stationery Office, 1984: 197. 11. Nottinghamshire Area Health Authority (T) (Planning Dept). 12. Clayton SG, Lewis TLT, Pinker G, eds. Obstetrics by ten teachers. London: Edward Arnold, 1980: 114-20. 13. Roberton NRC. A manual of neonatal intensive care London. Edward Arnold, 1981: 35-44. 14. Steiner ES, Sanders EM, Phillips ECK. Maddock CR. Very low birth weight children at school age: Comparison of neonatal management methods. Br Med J 1980; 281: 1237-40. 15. Vyas H, Milner AD. Other respiratory diseases in the neonate. In: Roberton NRC ed. Textbook of neonatology Edinburgh: Churchill Livingstone, 1986: 312-39. 16. Roberton NRC. The care of neonates with respiratory failure. In: Chiswick ML ed. Recent advances in perinatal medicine I. London: Churchill Livingstone, 1983: 179. 17. Jones RM, Rutter N, Cooper AC, Pullan CR Pneumothorax in the neonatal period. Anaesthesia 1983; 38: 948-52 18. Cooke RWI. In utero transfer to specialist centres Arch Dis Child 1983; 58: 483 -84. 19. Levene MI, Dubowitz LMS. Long term follow-up of low birth weight babies. Br J Hosp Med 1982, 28: 487-93
ROLE OF THE PHARMACEUTICAL INDUSTRY IN MAJOR CLINICAL TRIALS D. G. JULIAN J. R. HAMPTON Department of Medicine, University Hospital, Queen’s Medical Centre, Nottingham NG7 2UH; and Department of Cardiology,
University of Newcastle upon Tyne Mos’r large clinical trials
are financially supported by the pharmaceutical industry. Grant-giving bodies have limited funds and might well feel that projects other than expensive clinical trials will give a better academic return, and large trials are expensive. The industry perceives the financial
risk to be worthwhile because the returns from a successful trial can be high. Relations between investigators and the sponsoring company are usually harmonious but this is not always so, and a code of practice might prove helpful. The results of a trial must be credible if they are to be widely accepted and form a basis for clinical practice. In general the less the industry is seen to be involved, the more credible the trial will be. However, the drug company will be paying a large sum of money: to demonstrate a reduction in mortality from cardiovascular disease a budget of$1 million would be modest,$3 or$4 million is probably typical, and the most expensive of such trials has cost over$100 million. A company investing on this scale will be hesitant about handing over total control of a trial to the investigators. Furthermore the company may fear that’dull" data, needed for submission to regulatory authorities, may not be properly collected or analysed by the investigators. The problem is to balance the company’s desire to obtain what it has paid for against the investigators need for independence from
commercial pressure.
During the design phase the sponsoring company needs to be closely involved so that it can be assured that the trial will answer the question it is interested in. At this stage either the investigators the company can withdraw with no more than time wasted and feathers ruffled. The trial protocol must define in detail how the company will be involved in the running and analysis of the study, because it may be too late to prevent misunderstandings once the trial has begun. The company, for example, must retain the right to terminate the trial if it learns from other studies that the drug has undesirable side-effects, or if it becomes apparent that too few endpoints are occurring for the trial to be brought to a conclusion within an acceptable time. or
A crucial design decision concerns the place to which data will sent for recording on computer. The trial’s treatment code is most easily added when the information is computerised, so whoever has access to the data-recording system will (at least in theory) have access to the results as the trial proceeds. Investigators are well aware of the statistical dangers of "multiple looks" at
be
20. 21.
Eksmyr R, Eklund G. Early neonatal deaths in geographically defined populations with different organisation of medical care Acta Paediatr Scand 1985; 74: 848-54. Eksmyr R. Two geographically defined populations with different organisation of medical care. Comparison of perinatal risks. Acta Paediatr Scand 1985; 74:
855-60. 22. Barson AJ, Tasker M, Lieberman BA, Hiller VF. Impact of improved perinatal care on the causes of death. Arch Dis Child 1984; 59: 199-207. 23. Silverman WA. Retrolental fibroplasia: A modern parable. New York. Grune and Stratton, 1980: 28, 37-42, 69-89.
1259 the results of a trial in progress, but if this information is available to a company it might attempt to terminate the trial early, either because an apparently’good" result had appeared, which would be commercially beneficial, or because an apparently‘ ‘bad" trend suggested that money was being wasted.
Although most pharmaceutical companies behave properly in trials and towards investigators, there have been isolated examples of unethical conduct. For example, companies have tried to prevent publication of unfavourable results by pressure on investigators or on journals. One sponsoring company with access to data on a trial in progress found that the drug was proving ineffective, and so terminated the study; then the company used data from uncontrolled trials for advertising purposes. Another company used the results from a single centre in a multicentre trial to support its application for a drug licence, even though the results from the other centres were unfavourable. Most seasoned investigators have a personal blacklist of companies with whom they feel they cannot do business.
Investigators’ motives and actions are not always pure either: are plenty of examples of both academic and financial skulduggery perpetrated by trialists. The sponsor, like the investigators, has a right to expect honesty. It is thus important, both from the investigators’ and the company’spoint of view, that
there
the trial results should be held on a computer to which neither investigators nor company has access. It is also important to define in advance criteria for stopping a trial earlier than initially planned.
If the trial is to be seen to be immune to"multiple looks" by the investigators and free from interference by the sponsoring company, the data must be handled by an independent data centre and the results must be periodically reviewed by an ethical committee that is wholly independent of both the company, the investigators, and (so far as is
possible) the data centre itself. The data centre should probably be an academic unit with a reputation to maintain. It seems unavoidable that the company, even though it possesses adequate computing facilities, will have to fund the data
centre. The centre must agree with the company that data will be recorded in a way that is acceptable to the sponsor and which is compatible with the company’s own computer, so that at the end of the trial the sponsor can make any analysis it wishes. There may sometimes be room for compromise: for example, some of the routine data which are not directly important for the trial could be separated and retained in the company’s computer without the treatment code.
If the data handling problems are so large that the company’s computer must be used, then data should be available to the data centre for the intermediate analyses required by the ethical committee. The ethical committee must be independent, and it will only be seen to be so if it receives information from the data centre and not from the company. The ethical committee is the guardian of the trial. It protects the patients by ensuring that they are not exposed to harm nor deprived of benefit-ie, it applies the "stopping rules" if the results become unequivocal before the forecasted end. It protects the investigators by relieving them of the responsibility of unwittingly harming their patients or depriving them of benefit. And it also protects the sponsor by insisting on up-to-date information and by having the right to inspect the data centre if it has any reason to believe that data has not been properly collected. The ethical committee, whose composition must be agreed beforehand by the investigators and the sponsor, must insist on its independence from both parties once the trial has
begun.
Drug regulatory authorities usually hold the pharmaceutical company responsible for reporting adverse reactions to drugs (ADR). A company may therefore claim with some justification that it must hold the treatment code and receive all ADR reports. If an ADR is unexpected this is not important, but if, for example, we include the events expected in a cardiovafcular trial (such as death, myocardial infarction, and stroke) then it is clearly undesirable for the company to be privy to this information.
Ideally the ADRs who can supply to the company any "unexpected" ADRs associated with active treatment so that these can be added to the database the company will have built up from other studies. Such ADRs should be notified to the company quickly. "Expected" ADRs should, however, only be notified to the ethical committee, who should be responsible for informing the licensing authorities if any significant imbalance develops between treatment groups. This implies a delegation of responsibility from the licensing authority to the ethical committee; this has been permitted in some trials in the UK but is not official government policy. should be
sent to
the data
centre
Although the sponsor should not have access to trial data, it has an important role in ensuring that the study is properly run and the data properly collected. Non-medical trial monitors, employed by the company, will ensure continuity of drug supplies and any necessary technical facilities, and should visit the participating centres regularly to identify problems at an early stage. Because these monitors will have no access to the data, they will not be able to exert undesirable pressures; relations with the investigators can and should be close. The role of the monitors is especially important in international multicentre trials, because the international co-ordinator is unlikely to be familiar with foreign medical practices and will probably not know many of the investigators. However, even then the monitor must not be responsible for the trial conduct in his country; this must be a job for one of the investigators who will take on the role of national co-ordinator. Data must be sent through the national co-ordinator to the international co-ordinator and hence to the data centre, so that the co-ordinators at both levels can ensure a proper standard of recording. The final analysis must be done by the investigators independently of the sponsor. Although the company should have the right to comment on any manuscript, the ultimate decision on publication must be for the investigators and a journal editor. The results of all studies must be published, however briefly, because "pooling" (meta-analysis) is worthless if only positive trials are considered. To ensure that results are eventually available a register of trials could be established either with the regulatory authority or with a journal, and the publication of abstracts of protocols of major trials would alert investigators to seek the results if a publicaton did not appear. A journal editor will decide whether the design and conduct of a trial has been of a sufficiently high standard to merit publication. An editor will only feel confident that both the investigators and the sponsoring company have behaved with propriety if the appropriate organisation was built into the trial during its design. Correspondence should be addressed
to
K. R. H.