A partial defense of clinical equipoise

Abstract

In this essay, I suggest that a slightly modified version of Freedman’s formulation of the clinical equipoise requirement is justified. I begin this essay with a brief discussion of the equipoise requirement. In the second and third sections, I discuss several objections to the clinical equipoise requirement as well as two attempts to justify the equipoise requirement. In the fourth section, in order to gain some insight into what it means to say that a physician has an obligation to act in the best interest of her patient, I examine a similar obligation, namely, a parent’s obligation to act in her child’s best interest. I claim that there are several uncontroversial exceptions to this obligation, and these exceptions arise when a parent occupies a role more complex than the role of parent simpliciter. In the fifth section, I explore whether the exceptions to a parent’s obligation to act in her child’s best interest may shed light on some exceptions to a physician–researcher’s obligation to act in the best interest of her patient. These exceptions, I suggest, arise when a physician occupies a role more complex than physician simpliciter, and they ground a permission to conduct randomized clinical trials, even if doing so is not in the best interest of a physician–researcher’s patients. I furthermore claim that these exceptions provide justification for a modified formulation of the clinical equipoise requirement.

In 1974, Charles Fried, in Medical Experimentation: Personal Integrity and Social Policy, asserted that during all phases of a randomized clinical trial (hereafter RCT) a physician–researcher must be in a state of equipoise (1974: 143–153). Fried believed that the equipoise requirement, which dictates that physicians conducting RCTs must be in a state of genuine uncertainty concerning the relative efficacy of the different treatments being evaluated in an RCT, was a means of managing a dilemma associated with being a physician–researcher. Physicians are morally obligated to act in the best interest of their patients, and Fried explained that if a physician enrolls a patient in an RCT (or even suggests that her patient should enroll in an RCT) when equipoise is not present (when she believes that one of the treatment arms is more efficacious than the other arms), she is exposing her patient to the risk of being assigned to a treatment arm in which he will not receive what his physician believes is the best treatment. That said, researchers aim to conduct good science, and the RCT is considered the gold standard when it comes to determining the efficacy of different treatment options. Fried believed that the equipoise requirement would respect both of these goals in that it allows for RCTs, but only when conducting RCTs does not allow for the possibility that a patient will be randomized to an arm of the trial that the physician–researcher believes is inferior.

Subsequently, as I explain in more detail below, Benjamin Freedman (1987) proposed a modification to Fried’s equipoise requirement, which dictates that RCTs are ethically permissible if and only if there is disagreement within the clinical community concerning the relative efficacy of the different arms of an RCT. Although numerous essays criticizing the clinical equipoise requirement have been published, it is currently the dominant standard associated with whether it is morally permissible to conduct (or continue) an RCT (Brody et al., 2005: 1411–1414).

In this essay, I suggest that a slightly modified version of Freedman’s formulation of the equipoise requirement is justified. I begin this essay with a brief discussion of the equipoise requirement. In the second section, I discuss several objections to the clinical equipoise requirement. In the third section, in order to gain some insight into what it means to say that a physician has an obligation to act in the best interest of her patient, I examine a similar obligation, namely, a parent’s obligation to act in her child’s best interest. I claim that there are several uncontroversial exceptions to this obligation, and these exceptions arise when a parent occupies a role more complex than the role of parent simpliciter. In the fourth section, I explore whether the exceptions to a parent’s obligation to act in her child’s best interest may shed light on some exceptions to a physician–researcher’s obligation to act in the best interest of her patient. These exceptions, I suggest, arise when a physician occupies a role more complex than physician simpliciter, and they ground permission to conduct RCTs, even if doing so is not in the best interest of a physician–researcher’s patients. I furthermore claim that these exceptions provide justification for a modified formulation of the clinical equipoise requirement.

Theoretical equipoise and clinical equipoise

As explained above, Fried believed that the equipoise requirement was a means of resolving or managing the dilemma associated with being a physician–researcher. Fourteen years after Fried introduced the equipoise requirement, Freedman (1987: 142) argued that Fried’s equipoise requirement was far too stringent or too fragile and “poses a practical threat to clinical research” in that it rendered far too many RCTs unethical. Freedman states: “The reasoning or preliminary results that prompt the trial (and that may themselves be ethically mandatory) may jolt the investigator…out of equipoise before the trial begins” (1987: 141). And in those cases in which equipoise is present at the time of enrollment, it’s likely that data obtained before the trial is completed will break the physician–researcher’s state of equipoise, and the trial will have to be terminated prematurely.

Rather than outright rejecting Fried’s equipoise requirement, Freedman (1987: 143–145) proposed a reformulation of equipoise that he believed is less fragile than Fried’s formulation. He labeled his own formulation clinical equipoise and Fried’s formulation theoretical equipoise.

The primary difference between theoretical equipoise and clinical equipoise is that the former requires the physician–researcher to be in a state of uncertainty during all stages of an RCT, whereas the latter requires uncertainty within the expert clinical community. Freedman asserts (1987: 144) that clinical equipoise is present when “[t]here is no consensus within the expert clinical community about the comparative merits of the alternatives to be tested.” ¹ He believes that the clinical equipoise requirement will allow for more RCTs than does Fried’s theoretical equipoise requirement. ²

Objections to clinical equipoise

Freedman claims that if clinical equipoise is present during all stages of an RCT, the physician’s obligation to act in her patient’s best interest is not being violated. He states:

A state of clinical equipoise is consistent with a decided treatment preference on the part of the investigators. They must simply recognize that their less favored treatment is preferred by colleagues whom they consider to be responsible and competent. (1987: 144)

Essentially, Freedman claims that since medicine is a social practice (as opposed to an individual practice), the norms concerning appropriate medical care should be determined socially. In fact, medical negligence is typically predicated on the claim that a physician’s behavior fell below community standards. Hence, even if a physician–researcher believes that Treatment A, a new treatment, is superior to existing treatments, if there is no consensus within the medical community concerning the relative efficacy of Treatment A and existing treatments, the physician–researcher is not failing to do what is in the best interest of her patient if she conducts an RCT and her patient receives an existing treatment other than Treatment A.

As I have stated elsewhere, I am not convinced. When a patient selects a physician, he often does so because he believes that this physician’s judgment is best and/or because he expects this physician to suggest what the physician believes is the best treatment. Enrolling a patient in an RCT, and thereby letting chance determine whether this patient receives what his physician believes is the less efficacious treatment (even if there is no consensus concerning the relative efficacy of the different treatment arms), is inconsistent with providing one’s patient with what one believes is the best possible care (Gifford, 1995; Veatch, 2002).^3,4

A different line of criticism of the clinical equipoise requirement begins with the claim that the role of physician and the role of researcher are different, and not recognizing this difference leads theorists to incorrectly conclude that clinical equipoise must be present in order to conduct ethical RCTs. Howard Brody and Franklin Miller (2003a, 2003b, 2007), in several of the most important essays on this topic, explain that the “similarity position” holds that clinical researchers are bound by the ethical obligations associated with therapeutic medicine, while the “difference position” holds that therapeutic medicine and clinical research are distinct activities, each with their own set of ethical requirements. They claim that the difference principle is the best framework for understanding and managing the tension associated with being a physician–researcher.

Brody and Miller explain that successful research trials are predicated on obtaining generalizable knowledge, while therapeutic medicine aims at providing the best treatment for individualized patients. They claim that trying to reconcile these two activities is dishonest. “[T]he investigator cannot in good faith promise fidelity to do what is best medically for the patient-subject” (Brody and Miller, 2003a: 336). By accepting the difference principle and the obligations associated with being an ethical researcher, they assert, the researcher can maintain her professional and personal integrity. “[O]ne’s professional integrity is but one aspect of one’s entire human life, which demands that the person of integrity achieve a sense of moral wholeness and coherence among one’s total set of activities, goals and attitudes” (Brody and Miller, 2003a: 332). They furthermore claim that the duty of nonexploitation, which guides researchers, adequately protects research participants and makes sense of the duty of harm avoidance. They explain that the duty of nonexploitation dictates that “risks must be minimized, consistent with conducting scientifically valid research, and justified by the potential benefits to subjects (which may be lacking entirely) and by the potential value of the knowledge to be gained from the research” (2003a: 336).

A final objection to the clinical equipoise requirement is that it dictates that when a new treatment is extremely promising and there is a consensus among experts that the new treatment is superior to existing treatments, it is impermissible to conduct an RCT (even if doing so is necessary in order to obtain FDA approval), while a physician–researcher may conduct less promising RCTs. ⁵ This last is counter-intuitive at best. After all, the justification for conducting RCTs is that doing so has the potential to improve the medical treatment of future patients. One aspect of this is obtaining FDA approval for new treatments. Yet the clinical equipoise requirement makes this difficult or impossible when a new treatment is promising (when there is a consensus that this new treatment is better than existing treatments).

The physician’s obligation to act in the best interest of her patient

As Brody and Miller explain, the goals or obligations associated with the role of physician are different from those associated with the role of researcher. Researchers are concerned with developing better medical treatments or interventions that have the potential to increase the well-being of future patients, and this activity is best conceived of in terms of utilitarianism. ⁶ Physicians are obligated to do what is in the best interest of each of their patients, and this obligation is grounded in the duty of non-maleficence or benevolence. These different sources of obligation would not be problematic were it not for the fact that physicians frequently act as researchers. That is, utilitarianism, which guides researchers, allows one to conduct RCTs even if doing so is not in the best interest of one or more research subjects (in fact, utilitarianism requires this), if doing so will bring about the development of new medical treatments that in the long run will be better for society than current treatments. ⁷ Of course, physicians are prohibited from sacrificing the interest of any of their patients, even if doing so would maximize overall societal well-being. Hence, a physician–researcher seems to have the duty to act in her patient’s best interest, while at the same time she may be permitted (or even obligated) to sacrifice her patient’s best interest if doing so will provide medical knowledge that will, in the long run, increase (or bring about maximal) societal well-being.

Although bifurcating the roles of the physician–researcher may help us understand the moral/ethical tension associated with occupying this complex role, doing so unnecessarily complicates the matter in that it dictates that physician–researchers are bound by two irreconcilable sets of ethical requirements. Perhaps a better approach is to conceive of the role of physician–researcher as a complex but unified role. That is, perhaps physician–researchers are bound by a set of ethical obligations that are different from the obligations associated with being a physician simpliciter and those associated with being a researcher simpliciter. Just to be clear, like Brody and Miller, I am suggesting that we resist the temptation to assume that the physician–researcher is bound by the obligations associated with being a physician. However, contra Brody and Miller, I am suggesting that we resist the temptation to assume that the physician–researcher is bound by the obligations associated with being a researcher. Rather, we should attempt to determine what are the obligations of a unified moral agent who occupies a complex role with its own set of ethical obligations. (This last does not entail that physician–researchers are bound by the obligations of being a physician and the obligations of being a researcher.)

I hope to demonstrate that there are some situations in which a physician–researcher, as opposed to a physician simpliciter, is permitted to conduct RCTs, even if doing so is inconsistent with acting in the best interest of her patient. In order to explain why I believe this, it will be helpful to explore what common-sense morality says about the parent–child relationship. After all, a parent is said to have an obligation to act in her child’s best interest in the same way that a physician is said to have an obligation to act in her patient’s best interest. Yet there are many situations in which a parent is permitted or obligated to perform acts that are not in the best interest of one or more of her children.

The first exception to a parent’s obligation to act in the best interest of her child arises when a parent has more than one child. Parents frequently confront situations in which the interests of one child conflict with the interests of a second child, and it is impossible for a parent to act in the best interests of each of her children. For example, a parent may have two children whose piano recitals are scheduled for the same time in different locations, and the parent believes that attending each child’s recital is in the best interest of each of her children. Perhaps more interesting is the case of a parent who would like to have a second child. Presumably, it is morally permissible for this parent to perform actions that are not in the best interest of her one living child in an effort to increase the future well-being of her yet-to-be-born child. (In fact, it may be the case that having a second child is not in the best interest of her first child, yet this does not render having a second child immoral.)

A second exception to the obligation to act in the best interest of one’s child arises when a parent’s obligation to her child conflicts with her obligations to society-at-large. In these situations, a parent is sometimes permitted (or even obligated) to put societal interests (or the interests of other individuals or groups) in front of the interests of her own child. A parent is permitted (maybe even obligated), for example, to support a political candidate who plans to enact an inheritance tax law that will deny her child the opportunity to inherit enormous wealth. If the parent (correctly) believes that the law will significantly increase societal well-being but will not be in the best interest of her child (and her family as a whole), supporting the candidate and the law is permissible (or even obligatory).

A parent confronting a situation in which she cannot do what is in the best interest of her child and what is best for society, like the parent who cannot do what is in the best interest of both of her children, will likely be torn or pulled in different directions. On one hand, she wants to do what is best for her child. On the other hand, she is concerned about the well-being of other members of her society and wants to do what she can to increase their well-being. She may decide to support the candidate if she concludes that the amount of harm her child will suffer is not great and widespread benefits will result from the candidate being elected and the new inheritance law being enacted. More troubling for a parent is a situation in which she is contemplating supporting a politician who promises, if elected, to institute a universal military draft in order to increase the country’s military preparedness. Perhaps the parent agrees with the politician’s belief that a military intervention (a declaration of war) is the only means of stopping a tyrant from invading and conquering Europe and committing genocide. The parent may realize that if this politician is elected, her child may be drafted and suffer severe harm or even death. Again, I suggest that she will be pulled in different directions given her deep care (love) for her child and her concern for those who will suffer if the tyrant is not stopped.

One way to understand the latter exception (and maybe the first, as I explain below) is to recognize that the agent is not acting as a parent simpliciter or does not occupy the role of parent simpliciter. ⁸ That is, she occupies a complex role: that of parent–citizen. As a parent simpliciter, she may desire (or even be obligated) to do what is in her child’s best interest. But acting in the role of parent–citizen she is permitted (or even obligated) at times to do what is in society’s best interest, even if doing so is not in her child’s best interest. ⁹

The parent–citizen’s permission (or obligation) not to prioritize the well-being of her child is recognized by common-sense morality. As Michael Slote (1998: 184) explains in an essay discussing justice and the ethics of care, common-sense morality permits agents to favor or put more weight on their own projects. However, “common-sense also tells us that when the stakes are high enough” a moral shift takes place, and this permission is “displaced or superseded.” If, for example, an agent must choose between the death of a loved one or a stranger, she is (in most instances) permitted to choose to save the one she loves. However, if one person must die in order to save hundreds or thousands of people, say one’s child must die in order to foil a terrorist plot, the agent becomes less partialistic or even impartialistic. Similarly, Slote (1998: 184) explains that the common-sense permission to be partialistic is superseded when she occupies a political role and one’s choices in that role affect many people.

The first exception discussed above can be understood in the same way. There will be situations in which a parent has more than one child and cannot have an obligation to always act in the best interest of both of her children. This agent occupies a complex role, namely, the role of Parent of Child A–Parent of Child B (as opposed to occupying two distinct roles – Parent of Child A and Parent of Child B). Sometimes, while occupying this complex role, it will be impossible to act in the best interest of both of her children. Hence, this parent, who is not a parent simpliciter, is permitted to perform actions that are not in the best interest of each of her children. (As we will see in the following section, it is not crucial that the parent who cannot act in the best interest of both of her children be conceived of as acting in this complex role.) ¹⁰

The role of physician–researcher

I will now return to the physician’s obligation to act in the best interest of her patient. I hope to demonstrate that the obligation to act in the best interest of one’s patient is not always present when a physician occupies a more complex role than that of physician simpliciter. ¹¹ Put differently, the ethical requirements associated with being a physician may be different from those associated with being a physician–researcher or a physician–citizen.

Even though a physician has an obligation to act in the best interest of each individual patient, there will be instances in which the physician’s obligation to act in the best interest of one patient will conflict with her obligation to act in the best interest of one or more of her other patients. Consider two patients in need of surgery. Dr. Q believes that it would be in the best interest of each of these patients to have surgery the following morning. Unfortunately, there is only one time-slot available in the surgical ward of Hospital A and one time-slot available in the surgical ward of Hospital B, on the other side of town. In this case it will be impossible for Dr. Q to act in the best interest of both of her patients. Assuming the truth of “ought implies can,” Dr. Q must not have an obligation to act in the best interest of both of her patients.

Not only would it violate “ought implies can” to conclude that a physician who has more than one patient must act in the best interests of each of her patients, but this conclusion is inconsistent with common-sense ethics. Dr. Q, a psychiatrist, may have a patient, Patient A, who would benefit most if Dr. Q would spend several hours a day talking with him. However, common-sense ethics dictate that Dr. Q may have an obligation not to spend several hours a day talking with Patient A, even if not spending this time with Patient A is not in Patient A’s best interest, because Dr. Q has an obligation to spend time with other patients (or to make time to see new patients).

Notice how this exception to the physician’s obligation to act in the best interest of a patient is similar to the first exception to a parent’s obligation to act in the best interest of her child. In the same way that a parent who has more than one child is not always obligated to act in the best interest of each of her children, a physician who has more than one patient is not always obligated to act in the best interest of each of her patients. And in the same way that we can understand the exception to acting in Child A’s best interest by recognizing that the parent is no longer Parent of Child A simpliciter (rather, she is Parent of Child A–Parent of Child B), we can understand the exception to acting in the best interest of each of one’s patients by recognizing that the physician no longer occupies the role of Physician of Patient A simpliciter (rather, she occupies the role of Physician of Patient A–Physician of Patient B). While occupying this more complex role, the physician does not have the obligation to act in the best interest of each of her patients.

Physicians are also permitted to perform actions that contribute to the well-being of society-at-large (even if doing so is not in the best interest of an individual patient) in the same way that a parent may perform actions that are not in her child’s best interest but are best for society. Consider a situation like the anthrax scare after the 9/11 attacks. Let us imagine that there is a legitimate concern that some group is planning to launch a large-scale anthrax attack. Patient B, who has not been exposed to anthrax, requests that his physician, Dr. R, write him a prescription for Ciprofloxacin, an antibiotic prophylaxis used to treat anthrax exposure. Dr. R may believe that it is very unlikely that her patient will be exposed to anthrax (Patient B lives in a small rural town in Oklahoma), but Dr. R also believes Patient B would benefit (perhaps anxiety contributes to some underlying medical condition) if he had in his possession enough Ciprofloxacin to treat his whole family in the event of anthrax exposure. Nevertheless, it would be morally permissible (or even obligatory) for Dr. R to refuse to provide the requested prescription. Dr. R could justify her refusal by appealing to the potential shortage of Ciprofloxacin (should a widespread anthrax attack take place) and her obligation to society-at-large not to waste (or to be careful when administering) scarce medical resources.

Notice that this exception, like the prior exception, can be understood as a conflict that arises when the physician occupies a complex role. We might conceive of the agent as occupying the role of physician–citizen, in the same way that we can conceive of the parent, in the examples above, as occupying the role of parent–citizen. That is, there are some situations in which physicians (or parents) occupy a complex role, and the obligations associated with the role of physician simpliciter (or parent simpliciter) are different from the obligations associated with the role of physician–citizen (or parent–citizen).

Perhaps there are situations in which physicians occupying the role of physician–researcher may be permitted to conduct RCTs. Elsewhere, in an effort to determine what types of laws pertaining to criminal punishment might be enacted or supported by caring legislators, I explored punishment within the family. That is, I use the family as a heuristic or a model in an effort to gain insight into what laws or practices pertaining to criminal punishment might be just or ethical (Gelfand, 2004: 593–614). I would like to use a similar model in order to explore the issue of whether it is ever permissible for a physician–researcher to enroll a patient in an RCT. I believe this model will be useful because, as I explained above, like parents vis-à-vis their own children, physicians are said to have an obligation to act in the best interest of each of their patients. Let us begin by conceiving of a physician–researcher as a parent in a large family. We can then attempt to determine whether a caring parent–physician–researcher would ever enroll one of her own children in an RCT.

Before proceeding, I believe it is important to say something about the issue of paternalism. I want to emphasize that I am suggesting that we use the family as a model or a heuristic, and models, like metaphors, are limited in their scope. A model is not a duplication of reality. Rather, a model is a tool we can use to map one concept onto another concept in order to learn or illustrate something about the latter. Obviously, physicians, unlike parents vis-à-vis their own children, do not teach their patients manners or help them with their homework. And a patient, unlike a child vis-à-vis his parents, is not financially dependent on her physician. I hope to use this model in order to obtain some insights into what we mean when we say: “X has an obligation to do what is in Y’s best interest.” Since we use this phrase when describing both the obligation parents have to their own children and the obligation physicians have to their patients, the model may prove to be illuminating. That said, perhaps the problem of paternalism can be alleviated if, when we conceive of the parent–child relationship, we conceive of the relationship between parents and adult children. After all, parents are (or should be) far less paternalistic when interacting with their adult children than they are when interacting with their young children.

Let us imagine that a physician–researcher, Dr. P, is a parent in a large family, and one of Dr. P’s children has been diagnosed with Disease S, a rare and highly fatal disease. Dr. P became interested in Disease S when she learned that many of her relatives died as a result of contracting Disease S, which Dr. P discovered affects those with a rare gene mutation that is frequently found in her family’s DNA. Finally, let us imagine that Dr. P believes that a new treatment for Disease S, Treatment J, is more efficacious than existing treatments. Dr. P is contemplating conducting an RCT comparing the efficacy of Treatment J and other existing treatments. Dr. P knows that there are not many potential research subjects living in her city, and that she will, therefore, be faced with the question of whether to enroll her own child in this trial. She furthermore knows that if she enrolls her child in the trial, her child may not receive Treatment J. Would Dr. P, a caring parent, who is also a physician–researcher, decide to conduct and enroll her child in this RCT? Would Dr. P believe that she should be bound by the clinical equipoise requirement?

Dr. P is an expert and is aware of all of the existing experimental evidence concerning Disease S, Treatment J and other existing treatments. Since Dr. P believes that Treatment J is more efficacious than existing treatments, she would likely want her child to receive Treatment J and would be, therefore, disinclined to conduct the RCT. I suggest, however, that before making a final decision, if Dr. P is a thoughtful (and truly caring) parent, she would want to learn what other experts believe concerning the relative efficacy of Treatment J and existing treatments. ¹² After all, the stakes are huge, and Dr. P knows that researchers frequently believe a new treatment is superior to existing treatments, but after conducting RCTs they discover that the new treatment is inferior.

Empirical research on expert overconfidence supports my claim that physician–researchers should look to the expert clinical community before trying a new drug on a patient or conducting an RCT. In an essay that synthesizes the empirical findings of numerous studies pertaining to expert judgment and overconfidence, the authors conclude: “While expertise is often unrelated to accuracy, self-confidence, is not. Experts often proffer poor judgments but do so with hubris” (Cassidy and Buede, 2009: 455). ¹³ Physicians are not immune to this phenomenon. M. Podbregar and colleagues (2001) conducted a study in which physicians were asked to diagnose the cause of death of patients who died in the intensive care unit. Subsequently, autopsies were conducted on these patients in order to ascertain the true cause of death. Physicians who were “completely certain” that their diagnoses were correct were wrong 40% of the time. ¹⁴

These studies are instructive in that they clearly reveal that physicians, like other experts, tend to be overconfident. Hence, parent–physician–researchers who are considering providing a new treatment to their children (or other patients) or conducting an RCT should attempt to discover what other experts believe concerning the efficacy of this new treatment.

If Dr. P, who knows about expert overconfidence, discovers that there is no consensus within the expert clinical community concerning the relative efficacy of Treatment J and existing treatments, she would likely be torn or pulled in different directions when contemplating whether to conduct an RCT. She cares deeply for her child (and to a lesser degree her other patients) and believes that Treatment J is the best treatment, while she knows that some experts agree with her concerning the relative efficacy of Treatment J and other treatments and others disagree. Furthermore, she knows that in the future others (including her own relatives, maybe her grandchildren) will suffer from Disease S, and it is in their best interest for Dr. P to conduct an RCT and thereby obtain more evidence concerning the relative efficacy of the different treatments. For the sake of her own future patients (and the patients of physicians who do not believe Treatment J is superior to existing treatments) she will want to conduct an RCT. In addition, if Dr. P wants the FDA to approve the use of Treatment J for Disease S, which she expects will be in the best interest of those who will suffer from Disease S in the future, conducting an RCT will be helpful (or necessary). I am suggesting that parent–physician–researchers might arrive at different decisions in this situation. Some might conclude that conducting an RCT is the best option, while others might decide not to do so in order to ensure that their own children receive Treatment J. ¹⁵

If Dr. P discovers that there is a consensus that Treatment J is inferior to existing treatments, Dr. P would (or should) be reluctant or unwilling to conduct an RCT and thereby unnecessarily put at risk her own child and her other patients. After all, Dr. P knows about expert overconfidence and knows that many researchers in the past believed that a new treatment was superior to existing treatments and later discovered that the new treatment was worse than existing treatments.

Up to this point, we see that a physician–parent–researcher would (or should) act in accordance with the clinical equipoise requirement. If there is no consensus concerning the relative efficacy of Treatment J and existing treatments, they would consider conducting an RCT, but if there is a consensus that Treatment J is inferior to existing treatments, they would not conduct an RCT.

Would Dr. P, a parent–physician–researcher, conduct an RCT and enroll her child in the trial if she believes that there is a consensus among experts that Treatment J is superior to existing treatments? When first considering the possibility of conducting an RCT, Dr. P would want to ensure that her child received what she and other experts believe is the best treatment, and enrolling her child in an RCT would expose her child to the possibility of not receiving this treatment. Hence, she would be disinclined to conduct this RCT, and this is consistent with the clinical equipoise requirement.

However, in the same way that a parent might at first oppose an inheritance tax that would harm her children or a political candidate who is considering a universal draft in order to prepare for a large-scale war, after some consideration Dr. P might think differently. This is another situation in which a thoughtful, ethical agent will be torn or pulled in different directions. She will not want to expose her child to the possibility of not receiving what she and the expert clinical community believe is the best treatment, but she knows that her child is not the only person who will be affected by her decision.

Dr. P believes that Treatment J is better than existing treatments, and she believes that other experts agree with her. She knows that if she acquires additional experimental evidence, she will be in a better position in the future to evaluate the relative efficacy of Treatment J and existing treatments. Perhaps more importantly, she will obtain the evidence necessary for FDA approval. Put differently, conducting an RCT now, which Dr. P believes might not be in the best interest of some of those who currently suffer from Disease S, would be in the best interest of those who contract Disease S in the future (both her own patients and patients of other physicians). Dr. P, who must balance the interests of those who currently suffer from Disease S and the interests of those who will contract Disease S in the future, might feel like Agamemnon, who had to decide whether to sacrifice his daughter, Iphigenia, in order to carry out his duties as Commander of the United Greek Armed Forces or to sacrifice the interests of the United Greek Armed Forces in order to do what was best for his daughter.

I imagine that at times a parent–physician–researcher may want to ensure that her child receives Treatment J, while at other times she will want to conduct an RCT. (Or some parent–physician–researchers might choose to conduct RCTs, and others might not want to do so.) ¹⁶ Of course, if she decides to conduct an RCT in this situation, she will want to respect her patients and ensure, through the process of informed consent, that they are aware that they may not receive what she believes is the best treatment if they enroll in an RCT.

What can we learn from this hypothetical case? We can learn that if we are serious about creating conditions under which physician–researchers can develop new and better treatments while maintaining their personal and professional integrity, we should permit them to conduct RCTs in certain limited situations. More specifically, physician–researchers should be bound by the clinical equipoise requirement, except in those situations in which there is a consensus among experts in the clinical community that a new treatment is better than existing treatments. In these situations, they should be permitted (but not obligated) to conduct RCTs, even though doing so may not be in the best interest of their patient–research participants.

Just to be clear, the conclusion that physician–researchers should be permitted to conduct RCTs when there is a consensus that the new treatment is superior to existing treatments rests on the assumption that RCTs are necessary for FDA approval. If, as some have suggested is appropriate, the FDA changes its policy and no longer requires RCTs (perhaps the FDA will allow non-randomized clinical studies), the need for conducting an RCT may be eliminated and the necessity of conducting an RCT when there is a consensus that a new treatment is better than existing treatments may no longer be present. ¹⁷

Conclusion

Earlier in this essay, I discussed several objections to the clinical equipoise requirement. I explained that Brody and Miller (2003a, 2003b, 2007) claim that we err when we assume that physician–researchers are bound by the obligations of physicians. Brody and Miller suggest that in order to respect the integrity of physician–researchers, we must recognize that they are bound by the obligations associated with occupying the role of researcher, since research is the primary activity in which they are engaged. Requiring physician–researchers to claim that they are acting in accord with the duties associated with being a physician puts them in a position in which they must be dishonest to their patients (and even themselves). The second objection is that the clinical equipoise requirement renders the most promising trials unethical, while permitting RCTs when a treatment is less promising. This is counter-intuitive in that being prohibited from conducting RCTs on the most promising treatments makes it difficult or impossible to obtain the evidence necessary to obtain FDA approval for these treatments. Finally, I explain that Fried’s (1974) original equipoise requirement was motivated by the physician’s obligation to act in the best interest of her patient, yet Freedman’s (1987) clinical equipoise requirement allows a physician to enroll a patient in an RCT, even if she knows that doing so may result in a patient not receiving what she (the physician) believes is the best treatment.

In this essay, I take seriously Brody and Miller’s point concerning integrity. That is, like Brody and Miller, I claim that physician–researchers are engaged in a practice that is different from the practice of physicians simpliciter, and consequently, physician–researchers should not be bound by the obligations of physicians simpliciter. However, unlike Brody and Miller, I do not claim that physician–researchers should be bound by the obligations of researchers. Rather, I argue that physician–researchers occupy a complex role that is different from the role of researcher simpliciter and different from the role of physician simpliciter. I explain that physician–researchers are like parents who occupy a complex role (not merely the role of parent of one child simpliciter).

I furthermore conclude that in the same way that a parent–citizen is permitted (at times) to perform actions that are in the best interest of society-at-large (even though these actions are not in the best interest of her own child), a physician–researcher should be permitted (at times) to conduct RCTs (even though doing so might not be in the best interest of her patients). That is, physician–researchers should be permitted to conduct RCTs when there is no consensus among experts that one treatment arm is superior to other arms or when there is a consensus among experts that a new treatment is superior to existing treatments and conducting an RCT is necessary in order to obtain FDA approval for the new treatment. We see, therefore, that this approach takes seriously Brody and Miller’s argument for the difference principle and does not require physician–researchers to abandon the most promising RCTS (a weakness associated with Freedman’s clinical equipoise requirement).

The final objection discussed above is that the clinical equipoise requirement, which was initially formulated in order to respond to a problem with the theoretical equipoise requirement, does not respect the physician’s obligation to act in the best interest of her patients. Although this is also true of the approach I support in this essay, the approach I support does not try to minimize or deny that this is the case. Rather, I claim that a physician–researcher who is motivated by beneficence (and non-maleficence) should recognize that there is a tension between acting in the best interest of her patients and conducting an RCT. She wants to provide her current patients with the best treatments, and she wants to perform good research (which she expects will be in the best interest of future patients), but sometimes she cannot do both of these things. Deciding whether to conduct RCTs may (and perhaps should) cause physician–researchers to be confused or torn. This is consistent with a physician–researcher’s recognition that although she is not a physician simpliciter, she is not a researcher simpliciter either. Rather, she occupies the complex role of physician–researcher, and occupying this role allows her, when there is a consensus that a new treatment is superior to existing treatments, to perform actions that may not be in her patient’s best interest but are in the best interest of future patients, actions that would be impermissible were she acting in the role of physician simpliciter. Feeling or recognizing the moral tension in these situations does not imply that she is confused about the role she occupies. It does not imply that she is confusing her role as physician–researcher with that of physician simpliciter. Rather, feeling or recognizing the tension associated with being a physician - researcher is the natural response to being in a situation in which one has permission to perform two different actions, one of which is in the best interest of one’s patient (whose well-being the physician would like to improve) and the other of which may not be in the best interest of one’s patient but is in the best interest of society-at-large.