Zelen Randomization: Attitudes of Parents Participating in a Neonatal Clinical Trial
Introduction
The giving of informed consent is a prerequisite to participation in most randomized clinical trials (RCTs). In theory, it is appropriate and desirable that consent is given; in practice, informing a person fully about medical uncertainty and about the random allocation of treatment can be difficult and distressing for all parties involved. These inherent difficulties are further exacerbated in specialties where patients or their proxies are already experiencing a high degree of emotional strain. Discomfort with the process of consent can lead to poor rates of participation among professionals and therefore slow or inadequate rates of accrual of patients 1, 2, 3, 4, 5, 6. Fully informed consent in some circumstances has been described as “needlessly cruel,” with a statement that “attempts to gain the `informed’ participation of patients in randomized clinical trials are already doing harm in many individual cases [7].” Problems with consent therefore have serious implications for the success of RCTs, for professional morale, and also for the well-being of the participants.
Zelen addressed these difficulties by proposing a controversial departure from the conventional approach 8, 9, 10, 11, 12. Ordinarily, a clinician informs potential participants about a trial, and if consent is given, randomization follows. If consent is withheld, the standard treatment options (including no treatment) will usually be offered (Figure 1). In Zelen’s design, the clinician randomizes before asking for consent, if asking for consent at all. Consent, where sought, is for the randomly allocated treatment and for trial participation but not for randomization. This method has been termed variously randomized consent design, prerandomization design, and postrandomization design. In this article, we use the term “Zelen randomization.”
Two types of Zelen designs exist: “single consent” and “double consent.” In the single-consent design, the patients allocated to the control arm of the trial are not asked for consent—they simply receive the standard treatment without mention of the trial. Clinicians ask those allocated to the experimental arm of the trial whether they consent to receive the experimental treatment. If they decline, they may be receive the standard (or another) treatment (Figure 2). The double consent design differs in that the patients allocated to the control arm are asked for consent for the standard treatment and those among them who decline may receive the experimental (or some other) treatment. Those allocated to the experimental arm are asked for consent for experimental treatment. If they decline, they may receive standard (or some other) treatment (Figure 3).
Zelen’s designs are limited in that they are applicable only to certain trials. The single consent design is suitable only when the experimental treatment is compared with a best-known standard treatment. For a treatment to be considered experimental, it is unlikely to have been fully evaluated. So in this study, we concentrate on the situation in which the experimental treatment is only available within the trial, although the double consent design is suitable only when the experimental treatment is also available outside the trial.
Trials that use the Zelen approaches cannot be “blind” because the clinician is made aware of the treatment allocation to approach patients for consent to receive that specified treatment [13]. An assessment can, however, be made blind to allocation. Finally, trials requiring the active cooperation of the patient in a nonstandard protocol or follow-up are inappropriate when patients have not consented to participation.
Although there has been wide discussion of Zelen randomization, only a few trials are known to have used it, and most were conducted in the United States [for reviews, see 12, 14, 15]. Some use cluster randomizations [e.g., see 16], and we do not list these here. Most are in the field of cancer 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, but there are also examples from osteoarthritis [32] and neonatology 33, 34.
Zelen randomization has ethical and statistical advantages and disadvantages. Its appeal lies in the fact that in difficult circumstances a clinician need not ask for consent for the trial or for randomization, only for treatment. When stress levels are high, such simplification could ease both information giving and decision making. It might also improve rates of accrual, and therefore statistical power, in trials in which recruitment might be difficult. This might be because it leads to options that could be more acceptable to some patients and clinicians, but mainly because its design is weighted to include those who might otherwise have refused. In a conventional trial, all consenting patients are followed-up in an intention-to-treat-analysis; that is, they are analyzed in accordance with their allocation, regardless of whether they actually go on to receive the allocated treatment [35]. In Zelen’s designs, all randomized patients are followed on an intention-to-treat basis. All data are therefore appropriate for analysis as, in both of Zelen’s designs, trialists may analyze data even from those who decline the offered treatment.
Concerns about Zelen randomization arise from the fact that consent for randomization is not given in advance by any patients and that consent for treatment on protocol is not given by all. A proportion of the patients remain unaware of their role and are automatically included in the trial. The remainder have no explicit option to refuse to participate. Although it is widely accepted that patients whose data are used in epidemiologic studies are not necessarily informed of their roles in research, the acceptability of research without consent is less clear for RCTs. Of particular concern is the fact that not all patients are told that their treatment is chosen by chance 36, 37.
Moreover, there may be statistical concerns over bias. If, for instance, the experimental group of patients are required to attend for more follow-up than the control group, their relapses may be detected sooner. Lack of blinding may cause differential provision of concomitant treatments and/or differential recruitment into the two treatment arms. The largest statistical problem may be a reduction in power where an intention-to-treat analysis is used, the very concern that Zelen randomization aims to address. Power may be lost because of a dilution effect if a large proportion of patients reject their allocated treatment (termed crossover or transfer). Theoretical and empirical studies have shown that reduction in power resulting from crossover can be considerable, and the problem is compounded if the extent of dilution differs between the two trial arms 38, 39, 40. Because trialists cannot usually know the extent of crossover in advance, sample size calculations can be underestimated.
Although ethicists and statisticians have discussed the use of Zelen randomization, an important perspective is clearly missing from the debate: that of the participants themselves. Although studies have considered the views of healthy volunteers, [41] no evidence has yet demonstrated how those with experience of trials feel about the Zelen approach.
This article describes the responses to Zelen randomization from parents who had consented for their critically ill infants to participate in a conventional neonatal trial in which informed consent preceded random allocation of treatment. RCTs in such circumstances are of particular interest, because they inevitably involve persons who are extremely stressed. Parents are not only concerned for the well-being of their sick infants but are often within days or even hours of their birth and any associated traumas. They may be physically exhausted or deprived of sleep, and the mothers may be experiencing the effects of pain-killing or sedative drugs. Some have argued that consent for neonatal trials is not only difficult to obtain but may be inadequate, amounting to “an elaborate ritual” [42] or a “sham” [43]. In such a context, Zelen’s approach has been seen as appropriate [43].
Section snippets
The UK collaborative trial of extra corporeal membrane oxygenation
The UK Collaborative Trial of Extra Corporeal Membrane Oxygenation (The ECMO Trial) compared two methods of life support for mature neonates with acute but potentially reversible respiratory failure. The infants recruited to the trial were at high risk of death. At randomization they were already in a neonatal intensive care unit and receiving ventilatory support (“conventional management” or CM), one of the two methods the trial evaluated. Researchers compared CM to oxygenation of the blood
Discussion
Zelen randomization has been the subject of professional debate, but no attention has previously been paid to the views of those with actual experience of trial participation. The way in which parents in this study often engaged with and extended the issues involved in the subject was impressive and indicated the value of their particular perspective. Their interest was rooted in personal experience, and they empathized highly with others in similar situations. Despite the short time available
Conclusion
The subject of Zelen randomization proved to be a highly engaging topic for the parents in this study, who were divided in its acceptance or rejection. They saw a trade-off of protection and information, desirable to some, anathema to others. They also saw an opposition of the short-term benefits and long-term consequences. It became clear from their responses that Zelen randomization would create very different circumstances for participants depending on their allocation. Those who found this
Acknowledgements
This study was funded by the Department of Health and the Nuffield Foundation and developed from research funded by the Wellcome Trust. We thank Carole Harris from the ECMO Trial office and Sally Roberts at the Centre for Family Research for their help. Our thanks also go to the parents who kindly gave the information upon which this research is based.
References (49)
Integrating conflicting professional rolesphysician participation in randomized clinical trials
Soc Sci Med
(1992)Alternatives to classic randomized trials
Surg Clin North Am
(1981)- et al.
Randomised consent designs in cancer clinical trials
Eur J Cancer
(1995) - et al.
Routine formal fetal movement counting and risk of antepartum late death in normally-formed singletons
Lancet
(1989) - et al.
Randomised trial of monoclonal antibody for adjuvant therapy of resected Dukes’ C colorectal carcinoma. German Cancer Aid 17-1A Study Group
Lancet
(1994) - et al.
Informed versus randomized consent to clinical trials
Lancet
(1995) - et al.
Making sense of randomizationresponses of parents of critically ill babies to random allocation of treatment in a clinical trial
Soc Sci Med
(1997) - et al.
Oncologists’ reluctance to accrue patients onto clinical trialsan Illinois Cancer Center study
J Clin Oncol
(1991) - et al.
With womennew steps in research in midwifery
- et al.
Physicians’ reasons for not entering eligible patients in a randomized clinical trial of surgery for breast cancer
N Engl J Med
(1984)
Fundamental dilemmas of the randomized clinical trial processresults of a survey of the 1737 Eastern Cooperative Oncology Group investigators
J Clin Oncol
Randomized clinical trials in general practicelessons from a failure
BMJ
Fully informed consent can be needlessly cruel
BMJ
Statistical options in clinical trials
Semin Oncol
A new design for randomized clinical trials
N Engl J Med
Randomized consent designs for clinical trialsan update
Stat Med
Advantages and drawbacks of the Zelen design for randomized clinical trials
J Clin Pharmacol
Randomization designs in comparative clinical trials
N Engl J Med
Combined 5-fluorouracil and radiation therapy as a surgical adjuvant for poor prognosis gastric carcinoma
J Clin Oncol
Five year results of a randomised clinical trial comparing total mastectomy and segmental mastectomy with or without radiation in the treatment of breast cancer
N Engl J Med
Post operative adjuvant chemotherapy or radiation therapy for rectal cancerresults from NSAPB protocol C-01
J Natl Cancer Inst
Post operative adjuvant chemotherapy or radiation therapy for rectal cancerresults from NSAPB protocol R-01
J Natl Cancer Inst
Eight year results of a randomised clinical trial comparing total mastectomy and lumpectomy with or without irradiation in the treatment of breast cancer
N Engl J Med
Cited by (68)
The adapted Zelen was a feasible design to trial exercise in myeloma survivors
2020, Journal of Clinical EpidemiologyCitation Excerpt :Qualitative data from participants in a neonatal clinic elicited mixed attitudes toward Zelen randomization, with participants evenly split between accepting and not accepting the design. Some reported that they found the Zelen design underhand yet acknowledged if you were in the control group, it was kinder than an RCT, as you were not aware you had not received a potentially desirable treatment [26]. We used this design over a standard RCT because we felt it would cause patient distress if they were not allocated to the exercise intervention and may lead to contamination and participants to drop out.
OARSI Clinical Trials Recommendations: Design and conduct ofimplementation trials of interventions for osteoarthritis
2015, Osteoarthritis and CartilageCitation Excerpt :The value of courteously providing information to patients should not be underestimated, although it is possible that this results in either an increase in goodwill, or an increase in patient concern. Failure to provide this information risks psychological harm, as subsequent discovery of inclusion in an experiment might result in a sense of violation79,80. There are no clear rules for the degree to which such information should be given; this depends on the type of intervention studied and the potential to withdraw from the intervention or data collection, and should therefore be covenanted with patient boards, the professionals and ethical committees involved.
Conducting clinical trials in pediatrics
2009, Critical Care MedicineResearch in the Newborn Period
2020, Emerging Topics and Controversies in NeonatologyThe Value of Decompressive Craniectomy in Traumatic Brain Injury
2019, Evidence for Neurosurgery: Effective Procedures and Treatment