|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 23, No 19 (July 1), 2005: pp. 4468-4469 © 2005 American Society of Clinical Oncology. DOI: 10.1200/JCO.2005.01.2401
Withholding Information From Study Participants in Clinical Trials: Ethical ConsiderationsStony Brook University Medical Center, Stony Brook, NY To the Editor: I have read with interest the article by Retsas in The Art of Oncology section of the December 15, 2004, issue of the Journal of Clinical Oncology, as well as the accompanying rebuttal by Wieand and Murphy.1,2 Both articles raise valid points, but both neglect a relevant ethical question regarding equipoise and uncertainty: How does one justify the lack of information from an ongoing trial? Let us assume a randomized trial for regimen A compared with regimen B, and also assume it is agreed, on the basis of current knowledge, that neither can be considered superior. When any results become available from the trial, one of the regimens will more likely appear superior, at which point, ethical considerations arise regarding whether to advise patients and physicians of that information. For illustration, assume that analysis of the study reveals a borderline, but not quite statistically significant finding: that regimen A is more likely to cure than regimen B (P = .06). If the study is to be halted when some arbitrary probability is reached—say P < .05—what are the ethical responsibilities to patients randomly assigned to the inferior arm just before that limit is reached? The initial equipoise is severely compromised, and the degree of uncertainty is seriously reduced. In current practice, patients may be continued on study because treating physicians and patients are not made aware of ongoing study information, which is artificially unavailable by protocol rules that clearly favor obtaining a valid scientific result over full disclosure of information to study participants. If survival were the end point in the example above, how many patients would opt to be the last person to die to try to prove with P < .05 rather than P < .06 that regimen B really is inferior? From a stringent ethical point of view, the difference between strict equipoise and any change in probable outcome (eg, that based on any information derived from the study) is one of degree rather than principle. Even minimal differences in probable outcomes may be important to patients.3 Adaptive randomization schemes may improve, but do not eliminate ethical issues derived from withholding data.4 And yet, the higher we raise the P value, the less certain we can be that the results are correct. Randomized studies have clearly trumped observational studies in numerous areas; some documented by Drs Wieand and Murphy. It would seem that society, or the "far" (ie, the future) patient might benefit from knowledge of statistically valid study results, but the current study ("near") patient might clearly suffer in order to obtain that knowledge.5 This dilemma is not widely acknowledged, and should be addressed. There are many possible solutions to this specific problem concerning the broader area of societal versus individual benefits, but the issue should not be obfuscated by the currently widely accepted tactic of withholding information until the study has enough data to be scientifically valid. Currently, rules to prevent dissemination of knowledge to study investigators and subjects are common to randomized studies, to the possible detriment of the near patient. National Institutes of Health policy that regulates clinical trials since 1998 obligates a data safety monitoring board (DSMB) to "ensure ... the safety of the participants" and provide that "risks versus benefits are continually reassessed throughout the study period.6 These provisions must be reconciled with the coexisting DSMB obligation to ensure "the validity and integrity of the data." DSMBs live up to their names. Although there are clear imperatives for prevention of harm to patients, the safety of the data has become the paramount consideration, keeping information from others involved in the study to allow the generation of statistically acceptable results, perhaps at the expense of study patients. In the current age, there would seem to be little technological difficulty in informing all participants of results as soon the central reporting facility is notified. One could envision statistical information generated, graphed, and disseminated almost instantly on notification of an event. Most likely, the closer the study came to a definitive conclusion, the more difficult it might be to enroll subjects, thus compromising the study. Withholding information from study participants regarding the ongoing results of that study is a real ethical problem that should be acknowledged and openly debated. Author's Disclosures of Potential Conflicts of Interest The author indicated no potential conflicts of interest. REFERENCES
1. Retsas S: Treatment at random: The ultimate science or the betrayal of Hippocrates? J Clin Oncol 22:5005-5008, 2004
2. Wieand S, Murphy M: A commentary on treatment at random: The ultimate science or the betrayal of Hippocrates? J Clin Oncol 22:5009-5011, 2004 3. Ravdin PM, Siminoff IA, Harvey JA: Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 16:515-521, 1998[Abstract] 4. Karrison TG, Huo D, Chappell R: A group sequential, response-adaptive design for randomized clinical trials. Control Clin Trials 24:506-522, 2003[CrossRef][Medline]
5. Lilford RJ, Braunholtz D, Edwards S, et al: Monitoring clinical trials: Interim data should be publicly available. BMJ 323:441-442, 2001 6. NIH Policy for Data and Safety Monitoring, Released 1998. http://grants1.nih.gov/grants/guide/notice-files/not98-084.html
Related Reply
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2005 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
