Unblinding in Randomized Controlled Trials: A Research Ethics Case

Abstract

A pregnant woman was enrolled in a double-blind randomized controlled trial (RCT) in which participants were randomized to a placebo or a drug being tested to prevent a hypertensive complication. After completing the trial, the research participant insisted on being told which drug she received to prepare for a future pregnancy. This case highlights an element of RCT procedure that has received minimal attention—whether to unblind study participants at the end of their participation. Given that unblinding is not standard practice for nonserious adverse events, what actions are justifiable for the investigators to take? To synthesize the information about this case, we used the CASES model, created by the National Center for Ethics in Health Care to analyze ethics cases. Ethical principles that guide research emphasize communication with participants and the importance of reducing harm within the constraints of the scientific goals. Participants may value knowing which drug they received for future health care decision-making. We review information about the benefits and harms of unblinding.

Randomization was introduced in the 1920s as a tool to maximize the likelihood that varying individual characteristics that may affect outcomes are distributed evenly among conditions in experimental clinical trials. Outcomes of the study then can be attributed with greater confidence to the intervention under investigation.¹ For example, sociodemographic characteristics can influence health outcomes but, if randomization is successful, will be distributed more or less equally across the study arms. When that occurs, differences observed in outcomes between arms are less likely to be due to an association between demographic variables (such as socioeconomic status) and outcomes and more likely to represent a treatment effect.

Randomized controlled trials (RCTs) were originally developed for use in agricultural experiments. One of the first RCTs in a clinical setting was a study of streptomycin for tuberculosis in the United Kingdom in 1946. Randomization and allocation concealment were implemented for both participants and investigators involved in the study.² In psychiatric research, an early RCT focused on the efficacy of lithium for the treatment of mania at the Aarhus University Psychiatric Institute in Denmark in 1952.³ RCTs are now considered the gold standard in clinical trial design.⁴

Another frequent component of an RCT is blinding, or the concealment of group allocation from individuals involved in a clinical research study.⁵ Blinding minimizes bias, which is the risk that knowledge of the treatment assignment will influence both the subjects’ and assessor’s ratings of the outcome. It may also equalize the placebo effect across conditions. Double-blinded trials yield smaller—and presumably more accurate—estimates of treatment effect sizes than do non-double-blinded trials.⁶ In a single-blind study, the participant is unaware of the treatment allocation. In a double-blind trial, both the participant and all outcome evaluators are not aware of the randomization assignment. Double-blind trials are the most employed models for RCTs.⁷

A review of RCT unblinding indicated that only 45% of investigators informed either all or most participants of their treatment allocation after they completed the trial, while 55% did not inform any participants or informed only those who asked.⁸ The main reasons investigators did not unblind participants at the end of their participation were failure to consider this option (40%) and to avoid biasing results at study follow-up if the study was still ongoing (24%).⁹

Researchers agree that unblinding should occur for the patient’s safety when a serious adverse event (SAE) transpires and knowledge of the study drug could dictate the intervention to mitigate the health risk.¹⁰ To illustrate with a hypothetical example, consider a participant being treated for major depressive disorder in an RCT comparing the antidepressants sertraline and nortriptyline. She overdoses on the study medication and is taken to an emergency room. The physician contacts the principal investigator to determine the identity of the study drug, since the toxicity of the two drugs differs and the appropriate treatment depends upon drug identity. Tricyclics have cardiac toxicity, whereas sertraline toxicity can produce serotonin syndrome. These adverse reactions require different emergency medical treatments.

The issue of unblinding separate from the occurrence of an SAE has received less consideration. No standard practice exists for informing participants of their treatment allocation or the results of the trial.¹¹ Information about whether participants will be unblinded at any point is not a requirement for inclusion in research protocols submitted to institutional review boards (IRBs). The U.S. Department of Health and Human Services regulations (45 C.F.R. 46 and 21 C.F.R. 50 and 56, respectively) are silent regarding informing research subjects about individual treatment assignments or results. They neither directly require nor disallow disclosure.¹²

We present a case below to demonstrate the issues two of us experienced when a participant in an RCT requested to be told the identity of her drug treatment. Some details of the case have been changed to preserve anonymity. We used the National Center for Ethics in Health Care’s CASES model as the process for organizing the case discussion.¹³ The CASES model incorporates the following steps: clarify the consultation request, assemble the relevant information, synthesize the information, explain the synthesis, and support the consultation process. We reviewed publications about unblinding trial participants and details of the case to provide recommendations.

CLARIFY THE CONSULTATION REQUEST: CASE BACKGROUND

A pregnant woman was enrolled in a double-blind RCT in which participants were randomized in a one-to-one ratio to a placebo or to a drug being tested to prevent a severe hypertensive complication of pregnancy. The disease increases the risk for maternal and neonatal morbidity and mortality. No effective prophylactic therapy exists; hence, immediate delivery, usually preterm, is the indicated intervention.¹⁴

The participant had a complicated pregnancy history, as required for inclusion in this RCT. Her first pregnancy led to a very premature delivery with fetal demise. Her second and third pregnancies resulted in miscarriage. She participated in the RCT during her fourth pregnancy, which resulted in a healthy, full-term newborn with no obstetrical or neonatal complications.

After her participation in the study ended, she contacted the study team and insisted that she be told her drug assignment. As part of the initial consent process, the investigator explained the rationale for maintaining the blind during the study, which the woman accepted and apparently understood during her participation. She did not ask about unblinding during that time.

ASSEMBLE THE RELEVANT INFORMATION

The study was designed to evaluate safety for administration of the study medication to pregnant women at risk for severe hypertensive disorder. The drug being tested in this RCT was available by prescription for other indications not specific to pregnancy; however, obstetricians would be reluctant to prescribe it before the safety data were available from the study. The former participant stated that she deserved to have the choice about whether to search for a physician to prescribe it if her obstetrician would not. She also correctly stated that she was monitored closely for any negative outcomes during her participation in the RCT and, if she were taking the study drug, she tolerated it well and had a good outcome, which would be important for her prescriber to know. In her view, she was being disadvantaged by not knowing the identity of the study drug. Neither protocol nor consent form included any statement about postparticipation identification of the study drug’s identity.

The investigation included a placebo comparison to an active drug. A common assumption is that the placebo has no activity and no effect on the positive outcome. However, this view is not correct. Placebos are associated with substantial effects in medical studies.¹⁵ The placebo-treatment effect occurs in a psychosocial context. The placebo interacts with internal and external stimuli to change the functional neurobiology of the participant’s brain during the procedures inherent in an RCT. For example, the same brain mechanisms that are activated by drugs can also be activated by the placebo.¹⁶ “No drug” is not the same as “placebo drug.”

This case poses multiple questions. Given the information in the case presentation and that unblinding is not standard practice except in instances of SAEs, what actions are ethically justifiable for the investigators to consider?¹⁷ Does an individual have the right to find out her treatment allocation at the termination of her participation? Can a research team refuse to disclose a participant’s treatment allocation if the participant insists on knowing it? What are the ethics surrounding a research team’s decision-making to decide whether to disclose the identity of a participant’s treatment? Can investigators refuse to disclose the allocation if the participant’s physician believes that future clinical management would be compromised by not knowing the identity of the treatment? What are the steps that should be taken if the decision for unblinding is made?

To analyze this case, we reviewed medical facts, the patient’s preferences and interests, the sponsor’s preferences and interests, and ethical principles applicable to RCTs. Multiple databases, including PubMed, Medline, and Google Scholar, were searched for relevant articles in English. The Medline search (1980-2019) of “unblinding” and “treatment allocation” revealed 10 articles, “unmasking” and “treatment allocation” 4 articles, and “unblinding” and “ethics” 11 articles; the PubMed search (1980-2019) of “unblinding” and “treatment allocation” revealed 13 articles, “unmasking” and “treatment allocation” 4 articles, and “unblinding” and “ethics” 13 articles. The nine articles that discussed unblinding in double-blind RCTs were selected for review. Examples of excluded publications were those that described nonblind studies or maintaining the blind within studies. Additional articles were used to provide supporting evidence and clarify topics discussed in the selected articles. After reviewing basic ethics guidance, we will describe the findings of the literature review.

SYNTHESIZE THE INFORMATION

Ethical foundations.

Although standard practice in clinical trials has been not to unblind participants if no SAE occurs, failure to do so may conflict with the following generally held ethical formulations.¹⁸

The Declaration of Helsinki was created by the World Medical Association to serve as a guideline when conducting human subjects research. The Declaration stresses informed consent, confidentiality of data, vulnerable populations, requirements of protocols, a scientific rationale for conducting the study, and review of the study by an ethics committee.¹⁹ Although the Declaration notes that “[a]ll medical research subjects should be given the option of being informed about the general outcome and results of the study,” it does not specifically address release of information regarding study assignment.²⁰ However, the requirement that “sponsors, researchers and host country governments should make provisions for post-trial access for all participants who still need an intervention identified as beneficial in the trial” could be taken as a rationale for disclosure of the treatment received in cases in which benefit is likely.²¹

The principles described in The Belmont Report laid a foundation for regulations regarding the ethics of human subjects research in the United States.²² Among other effects, these principles obligate biomedical researchers to maximize benefits to the participants, within the constraints of the scientific aims of the study, while minimizing harm.²³ In relation to the case on which we are focused, the issue is whether avoidable harm was being done by not disclosing the study intervention postparticipation. Not disclosing the treatment intervention could risk harm for the individual and limit benefits, especially if the undisclosed treatment compromised future decision-making about the individual’s health.

The four principles of biomedical ethics identified by Beauchamp and Childress are frequently used to structure an ethical approach to research in health care.²⁴ These principles, which track those identified in The Belmont Report, are autonomy, nonmaleficence, beneficence, and justice. Respect for autonomy entails providing participants with sufficient information and time to understand the research. The principle of nonmaleficence includes the imperative to do no harm, while beneficence prioritizes optimizing the well-being of participants. Justice dictates the fair distribution of benefits and harms across populations of subjects.²⁵ With regard to the issue at hand, the principles of autonomy, nonmaleficence, and beneficence apply. Autonomy might have been compromised because the individual was not given information at the time of consent about the investigator’s plan regarding unblinding after participation in the trial. The withheld information could impact the individual’s future treatment, which might violate the principles of nonmaleficence, especially if she endured harm after participation, and beneficence, due to the priority for optimizing the well-being of the individual.

HIPAA privacy rule.

The Health Insurance Portability and Accountability Act’s Privacy Rule is applicable to clinical research in the U.S. conducted by covered entities (including almost all hospitals and most clinics) that involves access to or use of protected health information. With relevance to this case, under HIPAA, research participants have the right to access information generated in a research study if it is considered part of the institution’s designated record set. This set includes anything in a medical record generated by a HIPAA-covered entity and may include research records. One permitted exception to this rule applies to clinical trials; that is, the individual’s access rights are suspended while the trial is in progress if the participant provided consent to its being withheld prior to participation.²⁶ However, HIPAA requires disclosure on participant request after the trial is concluded. This was not included as a statement in the consent form in the case at hand.

Ethical implications of blinding and unblinding.

One consequence of unveiling the study allocation includes generating possible distrust and confusion for the former participant, especially if they were randomized to the placebo arm. Although potential participants are told during the consent discussion that they may receive a placebo, they nonetheless may believe that they will receive an active intervention, and the disclosure that they received a placebo may result in a loss of trust not only in the research team but also in their clinical relationships, especially if their treating clinicians are involved in the study.²⁷ Negative effects to unmasking also include the introduction of participant or investigator bias.²⁸ Unblinded participants could discuss their treatment and outcomes with subjects still involved in the trial or on social media. They may also change their perceptions of symptoms or change their behavior based on awareness of the treatment, which, in the event of long-term follow-up, can influence the outcome measures. They could reveal their treatment to the study investigators, which could bias the ongoing study. Study investigators may consciously or unconsciously view data, analyze outcomes, or report results differently.²⁹ Studies in which subjects become aware of their randomized assignment during the RCT due to inadequate blinding procedures have larger treatment effect sizes, which may not be valid.³⁰

There is less risk of bias in unblinding participants who complete their trial while the study is still in progress because the participant no longer provides data on the study outcomes.³¹ Moreover, the risk can be mitigated by inclusion of a nonblind medical monitor to undertake the disclosure.³² The monitor has no data collection or oversight role in the study and interacts with the team only about plans to meet with exiting subjects. This person meets with the participant to disclose the randomized treatment and debrief the participant.

An example of postcompletion unblinding was incorporated into the Prospective Study of Pravastatin in the Elderly at Risk study, a multicenter, randomized, double-blind RCT of the efficacy of pravastatin to prevent vascular disease. Participants were given the option of being unblinded to their treatment allocation after the completion of the entire study. The research team contacted them by letter to thank them for participating in the study, to offer a summary of the results, and to give them an opportunity to be unblinded. Participants were asked to indicate when a study nurse could contact them to be unblinded or given the option of not being contacted further. A postage-paid return envelope was provided. Participants who requested to be unblinded were contacted by a study nurse who informed them of their treatment allocation and cholesterol levels. A total of 2,067 participants were sent the information about the option of being unblinded, 1,492 (72.2%) participants responded, and 1,391 (93.2% of respondents) chose to be unblinded. No information was given about why participants declined to be unblinded. Many of the unblinded participants reported that they appreciated knowing their study medication allocation.³³ Such gratitude may result in participants having a positive attitude toward and trust in the investigators and research in general, which may lead to enrollment in future studies.

Unblinding procedures.

Unblinding participants so they can make an informed decision about further treatment may be advisable for several reasons.³⁴ Participants who knew the aggregate results of a trial but not their own treatment allocation reported being frustrated with the situation.³⁵ Participants may need support to decide whether they wish to know their treatment allocation, especially when the RCT reveals unexpected or unwelcome findings or when one group has much better results than the other.³⁶ Participants should understand the implications of being unblinded, especially regarding the knowledge that can reasonably be extrapolated from revealing the allocation. They should be told, for instance, that their outcomes, whether positive or negative, may have been due to factors associated with the trial experience other than the treatment modality itself.³⁷

In our case, the principal investigator discussed with the former participant the risks and possible harms of knowing which medication she received and the potential for disappointment. Potential outcomes were explored by asking her to imagine how she would feel if she learned that she had been assigned to receive the placebo? The study drug? The woman stated that knowing the drug she received during her successful pregnancy would be valuable clinical information in her planning for future pregnancies. If she was on active medication, she could request that treatment for a subsequent pregnancy. If she had received the placebo, she would know that she was able to achieve a successful pregnancy without the study drug (or any other pharmacologic intervention). She was hoping that she had been randomized to the placebo. If she had been treated with the study drug, she was aware that this trial was not testing the drug’s efficacy, which would be determined in a large-scale RCT. She stated that she would probably pursue being treated with the drug if she could find a prescriber, and she understood that she might not be able to get a prescription. If she were not unblinded, she felt that she would be harmed by not having information that could impact her future pregnancy planning.

The process of unblinding should be established early in the protocol planning process and included in the manual of procedures as well as in the consent process. The timing of the dissemination of the information should be optimized to benefit the participant.³⁸ We propose that it is the researcher’s responsibility to ensure that the participant understands how the research is being conducted, including whether there will be a process of unblinding at completion of participation and, if so, what this will entail.³⁹

EXPLAIN THE SYNTHESIS

The CASES model illustrates the many ethical issues that surround unblinding. Several themes emerged from the literature review. For unblinding due to an SAE, a nonblind medical monitor who has access to the randomization but does not participate directly in the study can unblind using a set of operationalized rules. This procedure preserves the blind for the study team during the RCT and permits unblinding without introducing bias to an ongoing study. A similar process can be used to unblind participants after participation in the study to maintain the blind for the research team.

Prior to unblinding the participant, the study team can discuss with this person the potential risks and benefits of knowing the treatment allocation. To address the potential distrust that may be introduced by unveiling the study allocation, the monitor can help participants understand the scientific rationale for randomization and use of a placebo and remind them of their agreement to be subject to this process in their consent to the study. To facilitate the transmission of the information to the participant’s clinicians, consent can be obtained to speak with them directly. Having a nonblind medical monitor provide support to a participant after the unblinding process can provide a sense of closure in the relationship between the participant and the study team.

Participants usually appreciate having the opportunity to know the identity of their study treatment. This courtesy may increase their positive perception of research and lead to a greater chance of their participating in future studies or influencing others to do so. Participating in RCTs is optional, and recruitment is a major challenge. Individuals who participate in research take risks to investigate new treatments for the benefit of the general population. Whether there will be an opportunity to be unblinded is an important element of informed consent.

SUPPORT THE CONSULTATION PROCESS: CASE OUTCOME

Following review of our case, the investigators requested that the randomized drug assignment be revealed to the participant. After consultation with the data safety and monitoring committee and the IRB, the decision was made to reveal the identity of her study drug. The nonblind study pharmacist released the identity of the drug to the participant privately, which allowed the rest of the study team to remain blind. The only information available to the pharmacist was the study identification number (as required by the IRB) to protect personal health information. Written consent was obtained from the participant to reveal her identity and contact information to the pharmacist. After learning of her treatment assignment, the participant expressed her appreciation to the research team.

We recommend that consent forms for RCTs include a clear statement about whether, when, and how unblinding will occur. Whether participants will have the choice to learn the blinded treatment to which they were randomized is important information for truly informed consent for the overall RCT. As illustrated by the case example, there may be times when information about treatment assignments is highly salient to participants because it may affect their subsequent medical care. Moreover, with proper procedures, randomized assignments can be unblinded without negatively affecting an ongoing study. Further research should be conducted in relation to the potential implications of unblinding participants. ♦