Discussion: View Thread

I have been approached by one of my colleagues, who is the site PI of a number of stroke clinical trials.  Some patients meet the inclusion criteria for more than one trial.  He is interested in not only the random allocation of treatment to patients, but also random allocation of patients to trials.  Is this necessary?  Your thoughts would be appreciated.

-------------------------------------------
Martin Durkin, MD, MPH
Palmetto Health Richland
-------------------------------------------

Would random allocation of patients to trials even be possible, considering that informed consent would be required from the patients before they could be included in any given trial?

-------------------------------------------
John Dawson
Postdoctoral Scholar
University of Alabama at Birmingham
-------------------------------------------

Each patient could be assigned randomly to a sequence of trials. Each one can then be offered enrollment in the first trial on the list. Those who accept have been randomized to a trial.

For patients who decline, you could quit there or you could offer each one the next trial in their individual sequence. They either accept or decline. 

Assignment to a treatment group in a trial is still random. Assignment to a trial itself is at least partly random if you have to offer 2 or more trials. 

If a patient declines to participate in 3 trials you may have evidence that they are not interested in research, but feel strongly about getting standard treatment. This many refusals might be useful information in itself.

If a lot of patients decline to participate in any particular trial, then you have evidence that the demands or risks of that trial are high, perhaps too high for most patients. This suggests that the treatment might not achieve a high degree of acceptability among patients unless it promises a high likelihood of success.

Offering patients two or more trials will probably lead to trials finishing sooner, which is probably good for everybody.

Randomizing patients to several trials doesn't even seem to be a lot of work, just additional record keeping. 

This seems worth discussing with IRB staff as having superior potential for patient benefit, if only because trials will finish sooner. I don't see why an IRB would not approve this meta-procedure. This randomization to trials does seem like human subjects research if records are kept for analysis and reporting. If no research on choice is planned, then it probably isn't even research. If you decide it is research (you decide if you plan to generalize the results, not the IRB) then you probably need to inform the patient that you are going to offer them the opportunity to participate in one of several randomly chosen trials, any of which seem suitable for their exact diagnosis and other circumstances and conditions, and that they are free to decline to participate after the trial is offered to them.

I think that initially, I would treat the process as not research, though I would probably still explain to a patient that they are being offered one trial among several that they are suited for. It seems to me that keeping the decision to offer one trial and not others a secret is unwise and shows a lack of respect for personal autonomy. Additional information and procedures shouldn't take more than a few minutes.

How many trials might a patient be eligible for?

Regards,



-------------------------------------------
David Smith
-------------------------------------------



Original Message:
Sent: 07-21-2014 12:19
From: John Dawson
Subject: Randomization of Patients to Clinical Trials

Would random allocation of patients to trials even be possible, considering that informed consent would be required from the patients before they could be included in any given trial?

-------------------------------------------
John Dawson
Postdoctoral Scholar
University of Alabama at Birmingham
-------------------------------------------

 

I would back up and first ask if there are any recruitment issues that may have selectively brought patients into the pool, where they believed they were volunteering for Study X.  If upon assessment you discover that they are eligible not only for Study X but also Y & Z then I would probably opt for keeping them in Study X unless they explicitly agree to be considered for the expanded set of studies.  This, however, is more of a transparency issue than strictly speaking a study design issue.

Assuming that there are not any of these sorts of constraints, and that informed consent is obtained for all studies, then I would randomly assign to the study and then to the group.



-------------------------------------------
David Mangen
-------------------------------------------

This seems like a very bad idea to me. There are no statistical advantages to this extra layer of randomization. The population being recruited to a clinical trial is a convenience sample and randomizing this convenience sample across several studies does not change this fact.

There is a disregard for patient autonomy, as another poster noted. If a volunteer likes one study more than another, why would you take that choice away from them? They're more likely to be compliant with a trial that they like more.

There's the extra layer of complexity, and you have to decide how to handle patients who decide after randomization to Trial A, that they do not like the trial after reading the informed consent form. Do you re-randomize them among the remaining studies? What do you do if they read the informed consent form for Trial B and decide that they really liked Trial A better?

There is only one thing I can think of saying in favor of randomizing across trials. It helps assure that all researchers get an equitable distribution of patients into their own individual trials.

Given the many potential problems, I would avoid randomizing across trials. If a patient is eligible for more than one trial, tell them that fact and let them make the choice.

-------------------------------------------
Stephen Simon
Independent Statistical Consultant
P. Mean Consulting
-------------------------------------------

Randomization to one of the eligible trials is fine in principle, but I wouldn't encourage it in practice unless I knew the investigator was really capable of randomizing. Even an investigator who appreciates the abstract distinction between random and haphazard may not be aware that time and resources (e.g., a statistician or reputable table of random numbers) are needed to plan and execute a process that's better than haphazard. Calling an assignment to study random when it isn't would dilute the importance of that concept, and that would be unfortunate given that the time and resources to do it right are already being invested where it matters more, assignment of subjects to treatment. Let the investigator to use his or her best judgment as to which trial the patient is best suited for, and if the studies are more or less equal in that regard, assign the patient however he or she likes. Just don't call it random unless it really is. 

-------------------------------------------
Richard Bittman
President
Bittman Biostat, Inc.
-------------------------------------------



Original Message:
Sent: 07-21-2014 14:39
From: Stephen Simon
Subject: Randomization of Patients to Clinical Trials

This seems like a very bad idea to me. There are no statistical advantages to this extra layer of randomization. The population being recruited to a clinical trial is a convenience sample and randomizing this convenience sample across several studies does not change this fact.

There is a disregard for patient autonomy, as another poster noted. If a volunteer likes one study more than another, why would you take that choice away from them? They're more likely to be compliant with a trial that they like more.

There's the extra layer of complexity, and you have to decide how to handle patients who decide after randomization to Trial A, that they do not like the trial after reading the informed consent form. Do you re-randomize them among the remaining studies? What do you do if they read the informed consent form for Trial B and decide that they really liked Trial A better?

There is only one thing I can think of saying in favor of randomizing across trials. It helps assure that all researchers get an equitable distribution of patients into their own individual trials.

Given the many potential problems, I would avoid randomizing across trials. If a patient is eligible for more than one trial, tell them that fact and let them make the choice.

-------------------------------------------
Stephen Simon
Independent Statistical Consultant
P. Mean Consulting
-------------------------------------------

In my opinion, it is not necessary to randomly allocate patients to trials. Random allocation of treatments to patients as they enter a given  trial is sufficient.
It appears that this colleague who wants to randomize only some patients who meet the inclusion/exclusion criteria of some but not all trials, does not have clear idea of "Randomization". I am saying this because not all patients will be eligible for randomization to all the planned trials. Different number of patients will be eligible for randomization to different number of trials. Some patients could be eligible for randomization to a single trial.So each patient will have different chance of getting  randomized to a given trial. It is clearly not acceptable. Further, it is not clear whether these trials have identical consent forms! If not, a patient randomized to a particular trial may refuse to sign it, even if he/she is eligible to enter another trial and could sign the consent form of that particular trial. 
It is unclear what does this colleague hope to achieve by random allocation of only some patients to some trials. As I see it, it will introduce lot of complexity in the analysis and perhaps confusion in the interpretation of the final conclusions.
-------------------------------------------
Sheela Talwalker, Ph.D.
CSO, T'Walker Consulting
-------------------------------------------