Research Versus Care

This tension between research and care is ever present in public health emergencies. The urgent desire to help current patients with whatever is available may appear to be in conflict with the need to learn as much as possible about potential interventions in order to help current and future patients. During the Ebola epidemic, some caregivers may have felt that providing clinical care and conducting clinical research were mutually exclusive and that one could not be done without harming the other effort. Clinical research and clinical care are sometimes at odds because care focuses on the individual, current needs of a specific patient (Sacristán, 2015), while clinical research benefits future patients and not necessarily the specific patient enrolled in the research; however, patients who enroll in clinical trials often benefit from receiving better medical care than patients not enrolled in trials. Furthermore, in a research setting health care decisions are not based only on the interaction of one health professional and one patient, but are often controlled through the process of randomization, and adherence to a standard protocol is required. This lack of autonomy on the part of both clinician and patient can add to the tension.

Despite tensions between research and clinical care, they can also be seen as two sides of the same coin, ideally conducted in tandem (Sacristán, 2015). The Declaration of Helsinki addresses this in its guidelines for physicians; it states that research and medical care may be combined “only to the extent that the research is justified by its potential preventive, diagnostic, or therapeutic value and if the physician has good reason to believe that participation in the research study will not adversely affect the health of the patients who serve as research subjects” (WMA, 2013). In the initial stages of an outbreak or epidemic, when care centers and workers are inundated with patients, it may be appropriate to focus the limited resources available on clinical care so as not to detract from the clinical response and incorporate research later when staff is not spread as thin. However, it is critical that the possibility of integrating research into clinical care is part of the discussions and planning from the outset, because it can take considerable time to obtain necessary approvals for research and to fully train the study team, while the opportunity to initiate a trial is continually reassessed as conditions on the ground change. If the planning does not occur early in the epidemic, the epidemic may be waning—or over—by the time a study is designed, approved, and ready to be implemented. This aptly describes the Ebola outbreak in the first 6 months of 2014.

An initial question was whether it was ethical to conduct research at all, given the extreme demands already placed on frontline care providers during the epidemic. Because there were no drugs or vaccines approved to treat or prevent Ebola or ready to enter into clinical trials at the outset of the epidemic, many felt that there was an ethical imperative to conduct such research as quickly and safely as possible. Providers needed to learn how best to treat patients or prevent new Ebola infections and to assess how health systems could be configured and equipped to meet these health needs. The WHO Ethics Working Group report from October 2014 stated that there was an ethical obligation to do research during the epidemic and that research should be part of the public health response (WHO, 2014b). Only by increasing the knowledge base about Ebola and about the merits of various treatment or prevention strategies could clinicians be sure that their efforts improved patient outcomes and communities be reassured that their scarce resources were used wisely and efficiently (UNESCO, 2006).

Expanded Access

Although providing experimental therapies in the context of a clinical trial is the ideal way to monitor and minimize risks of unproven agents while maximizing the scientific information gained, in some circumstances it is appropriate to administer unproven agents outside of an approved clinical trial. This is called an expanded access exemption or “compassionate use.” In the United States, a number of conditions must be met in order for a patient to be granted access to a drug under expanded access: (1) there is no comparable or satisfactory therapy available, (2) the probable risk from the investigational product is not greater than the probable risk from the disease, and (3) providing the investigational product will not interfere with the conduct of clinical trials (FDA, 2016).

The foreign aid workers, like Brantley and Writebol, received experimental therapies under an expanded access framework. In October 2014, the WHO working group report referred to compassionate use of investigational products as justifiable as long as data are collected and shared (WHO, 2015). However, most of the examples of expanded access provided were cases of foreign health workers who were evacuated from West Africa to the United States or Europe in order to ensure that they received optimal supportive care, and who, in desperation, were also offered whatever experimental intervention was available—and more than one if available (Enserink, 2014b). In this context, it would have been extremely difficult to attribute either beneficial or detrimental outcomes to any one of these investigational agents. The use of investigational agents under expanded access in these situations did not contribute to the knowledge base, but they did serve to initiate rumors that there was a cure for the foreigners that was not being made available to Africans. The belief that investigational agents in the very early stages of development were likely to be highly effective furthered the view that randomized controlled trials were unethical. For example, Caplan et al. concluded that because “all available agents have been variously deployed against infected persons treated in the United States and Europe, the case for randomization to placebo in West Africa is morally suspect” (Caplan et al., 2015, p. 6).

The assertion of a right to access an intervention without established efficacy in these circumstances is controversial at best. Any such right must be grounded in a concern for individual health, and there is no evidence that investigational products in the early stage of development will promote the health of humans treated with the agent. In fact, most agents in the early stages of development are eventually proven to be ineffective or even potentially harmful (Dresser, 2009). Additionally, a right to access interventions approved for treating another condition, but without established efficacy for another particular condition might be unbounded in its scope since there are potentially a great many interventions that have no evidence of efficacy for a particular condition but whose use might be supported on theoretical or even speculative grounds. Moreover, rights such as this do not exist in a vacuum. They can only be honored through the expenditure of time and resources for research. When limited resources have to be used to address the health needs of many individuals, proof of efficacy is a reasonable requirement for the use of resources. Absence of evidence of efficacy thus reduces the strength of the claim that scarce resources should be directed to the provision of novel interventions of unknown value.

Others argued that expanded access should be avoided because its “use exposes many patients to investigational interventions, often undermines fair access to experimental agents, compromises the collection of robust data to determine the safety and efficacy of interventions, and consumes scarce resources for uncertain clinical benefits” (Rid and Emanuel, 2014, p. 1844). Additionally, given the limited supply of experimental Ebola treatments and vaccines at the time, randomized trials may actually have been the most equitable way to distribute these products (Largent, 2016; Rettner, 2014). The strongest argument for providing expanded access to unproven therapies during the Ebola epidemic is that the high lethality of the disease tipped the ethical scales in favor of providing interventions that could be helpful, however remote that prospect of benefit may have been and even given the potential for harm. This argument springs from the principle of beneficence—the notion that medical care providers should seek to help patients. Yet even under such conditions, the social costs of providing expanded access merit consideration. Specifically, under circumstances like the Ebola epidemic, the principle of beneficence supports providing products under an expanded access exception when the following conditions are met (Darrow et al., 2015; FDA, 2016):

• A sufficient amount of the product is available after supplying the needs of clinical trials.
• Providing expanded access would not preclude or delay the initiation of more conclusive investigations of the intervention in properly designed studies. This could occur, for example, if the availability of investigational products off protocol depleted the supply of individuals willing to enroll in studies that could yield generalizable knowledge about the product's safety and efficacy.
• Existing evidence does not suggest such an unfavorable risk–benefit balance that the product would not even “make the cut” for inclusion in clinical trials.

Conclusion 2-1 The use of unproven experimental therapies—especially those in the early phases of drug development—under an expanded access exemption to patients regardless of nationality or where they are located, not only fails to provide information on safety or efficacy, but also creates inequities with the larger affected population during an epidemic. Such uses can promote the public misconception that a safe and effective treatment exists and may generate mistrust of researchers and research efforts that will make it more difficult to launch clinical trials when additional interventions become available.

Identifying Candidates to Research

To identify candidates for trials, researchers looked to the few Ebola-specific agents that were in early stages of development, and also explored the possibility of repurposing approved drugs for the new indication of Ebola. Although already approved for another indication, repurposed drugs would still require clinical investigation in order to determine the efficacy and safety of the drug for the new indication and patient population. Researchers sought to investigate numerous agents with limited evidence of potential value in the search for a highly efficacious medicinal product to treat Ebola. This pursuit resulted in a glut of proposals that “flooded the in-boxes of staff at the WHO and research funding agencies. Silver nanoparticles. Cholesterol-controlling statins. A breast cancer drug. Intravenous ozone. Vulture gastric fluids. An influenza antiviral. Interferon. Almost anything you can think of [was] being advocated as a potential Ebola curative, often with few or no data to support the case” (Branswell, 2014). In order to prioritize and select compounds to study in clinical trials for the treatment of Ebola, the WHO convened the Scientific and Technical Advisory Committee on Emergency Ebola Interventions (STAC-EE) in November 2014 in Geneva. Here the committee noted their inundation with an ever-increasing number of potential agents for proposed trials: “WHO and partners receive daily proposals for potential products against [Ebola] from the scientific community” (WHO, 2014d).

In order to winnow these down, the STAC-EE committee developed a set of criteria regarding the minimum levels of evidence required for an agent to be considered for clinical trials. The summary of this meeting makes clear that there was disagreement about the relative importance of the availability of an agent and the efficacy of that agent. Some participants believed that products that were readily available or easy to produce, such as brincidofovir and favipiravir, should be prioritized despite a lack of evidence that they would be effective in Ebola patients. Others thought that “availability was not a reason to study drugs with weak supporting data” and favored prioritizing drugs that had shown strong preclinical evidence of efficacy, such as monoclonal antibodies and small inhibitory RNA, even if they were less readily available (WHO, 2014d). Ultimately, the STAC-EE committee published a list of around 20 potential agents, acknowledging that “many of these have already been tested and shown to have no activity against the virus” (WHO, 2014d). Participants said that scientists and developers should “assess themselves whether further investigation is warranted” (WHO, 2014d); others later said that the list was not helpful, as it showed products “that barely worked in a mouse . . . in the same column as something that was shown in a non-human primate to be very effective.”¹ This lack of prioritization of the agents to be tested may have reduced the likelihood that clinical trials would identify beneficial agents. If fewer agents were proposed it is possible that more comprehensive data would have been available on the more promising agents.

It is difficult to estimate how long an emergent epidemic will last or how many people will be affected, and therefore difficult to determine the number of people who could enroll in a trial. By limiting the number of agents studied and the number of trials allowed to proceed, the trials that do proceed will be more likely to enroll enough participants to reach conclusive results, and the likelihood of identifying effective interventions will be maximized. In order to achieve this prioritization and limiting of trials, a rapid response body that offers access to broad expertise and mechanisms to avoid conflicts of interest in decision making is needed. Such a body should have the ability to convene the expert members at short notice, and have the authority to determine which studies will actually proceed (see Chapter 7 for further discussion).

Conclusion 2-2 In the event of a rapidly progressing outbreak it is critical to create a mechanism to prioritize investigational agents for study and limit the conduct of the clinical trials to a small number of products, focusing on those with the most promising preclinical or human clinical data, in order to maximize the likelihood that meaningful results will be generated.

Choosing Appropriate Trial Designs

Trial design was one of the most contentious areas of debate among those participating in discussions about Ebola clinical trials. Stakeholders disagreed about the proper approach to ethical, scientific, and practical issues, and they disagreed about how these issues should inform design decisions. (A synopsis of some study designs is presented in a table in Appendix B.) The clash between humanitarian medicine and research science was also evident in these discussions. As one representative present at the meetings later offered, “The fundamental tension is between the obligation to treat patients with whatever intervention offers the best hope of success and the obligation to gather objective evidence in a scientifically rigorous manner. The stakes are high in a crisis in which time is short and consequences of treatment failure are deadly” (Dawson, 2015, p. 45). Stakeholders struggled with issues such as using randomized trials versus nonrandomized alternative designs, the use of a standard-of-care control arm, and the fair distribution of limited product. Though stakeholders disagreed on all these topics, they later were in agreement on one issue: Too much time was spent debating trial design, rather than quickly implementing trials and discovering safe and effective products in time to fight the epidemic² (Cohen and Kupferschmidt, 2014).

Randomized Controlled Trials

The gold standard for a clinical trial continues to be the randomized controlled trial (RCT). An RCT is considered the best tool for assessing the efficacy of a treatment and is used for several reasons: it avoids selection bias, improves comparability between the experimental and control arms, and allows for valid statistical testing that permits a reliable assessment of the likelihood that observed differences in outcomes between arms could be due to chance (Suresh, 2011). In an RCT, patients are allocated by chance to an arm of the study. There are several types of arms, including (NIH, 2016)

• Experimental: A group of participants who receive the intervention that is the focus of the study, that is, the investigational treatment or vaccine.
• Control: A group of participants who do not receive the investigational treatment or vaccine.
  • Active: A group of participants who receive an intervention that is considered to be effective but that is not the investigational treatment (e.g., a vaccine for hepatitis rather than the product under investigation).
  • Placebo: A substance that does not contain active ingredients and is made to be physically indistinguishable from (i.e., it looks and tastes identical to) the investigational treatment or vaccine.
  • Sham: A procedure or device that is made to be indistinguishable from the actual procedure or device being studied but that does not contain active processes or components.
• Standard of care: A group of participants who receive standard medical care for the condition being studied. Standard of care is sometimes but not always provided in conjunction with the experimental treatment, a placebo, or sham.

While most trials employ randomization through the comparison of two treatment approaches (e.g., the investigational therapy versus a control), others may compare multiple approaches, with or without a control group. For example, a factorial design randomizes each individual to two or more treatments over the course of a single trial so that multiple questions can be addressed in a single trial (Montgomery et al., 2003). When two or more known effective treatments are available, a head-to-head comparison of treatments (without a control) can be an appropriate design to compare their relative effectiveness and safety. Multiarm, multistage designs and Bayesian adaptive platform designs have also been proposed for situations in which multiple experimental agents are simultaneously available (Gurrin et al., 2000; MRC CTU, 2014). “Adaptive clinical trials are designed to take advantage of accumulating information, by allowing modification to key trial parameters in response to accumulating information and according to predefined rules” (Lewis, 2012, screen 5). (See Box 2-1 for more information on adaptive trial design.) Each of these trial designs involves randomization. There are many procedures for the random assignment of participants to treatment groups in clinical trials. Simple randomization (i.e., the investigational therapy versus a control) is just one form of randomization—some others include block, stratified, and covariate adaptive randomization—each type has its advantages and disadvantages for a given situation (large versus small trials, known prognostic factors, etc.). Clinical trial teams will need to assess the context surrounding the trial before determining which type of randomization to use; however, the benefits of randomization (as discussed above) contribute to it being an essential part of clinical trials to establish efficacy.

BOX 2-1

Adaptive Trial Design.

It should also be noted that the analysis of trial results depends in part on the design chosen for the trial. Because patient enrollment for a clinical trial is typically staggered, the regular interim analysis of trial results allows rapid identification of highly effective (or harmful) treatments, enabling researchers to terminate a study early if a treatment appears particularly beneficial (or harmful). Several statistical approaches to interim monitoring, constructed to avoid increasing the false positive rate, are widely used, as discussed in Proschan et al. (2006) (see Box 2-2 for additional information on statistical guidelines for early termination of a trial).

BOX 2-2

Statistical Guidelines for Early Termination of Clinical Trials.

Ethics in Human Subjects Research

Since the promulgation of the Nuremburg Code in 1947, numerous efforts have been made by different organizations to codify the basic ethical principles that should govern research with human subjects (CIOMS, 2016; COE, 1997; HHS, 2009; National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, 1979; Nuremberg Code [1947], 1996; UNESCO, 2006; WMA, 2013). A vast scholarly literature has emerged over time on the central ethical questions, as new situations have been identified and thinking has evolved. The committee has identified seven moral requirements that are widely recognized in authoritative guidance documents and the scholarly literature that are of particular importance for evaluating trials conducted during the Ebola epidemic and other similar circumstances in the future, these requirements are (1) scientific and social value, (2) respect for persons, (3) community engagement, (4) concern for participant welfare and interests, (5) favorable risk–benefit balance, (6) justice in the distribution of benefits and burdens, and (7) post-trial access (see Box 2-3 and Appendix C.)

BOX 2-3

Moral Framework for Research.

6. Justice in the Distribution of Benefits and Burdens

To be ethically permissible, the benefits and burdens of research must be fairly distributed. Research should not focus disproportionately on the health needs of some groups while neglecting the health needs of others, and the burdens of research participation should not be borne solely by groups of people who are unlikely to benefit from the knowledge generated. (See Box 2-4 for a discussion on the inclusion of pregnant women and children in clinical trials.) Some groups are particularly vulnerable to neglect or exploitation because of deprivation, disease, marginalization, or oppression. Thus, fairness requires that these groups not be excluded from research nor should they bear a disproportionate share of the burdens of research participation (CIOMS, 2016). Victims of public health emergencies are placed at increased risk and heightened vulnerability in many ways, but also may have unique health needs that cannot be studied outside of the emergency situation. Failing to conduct research in such situations under the guise of protecting the vulnerable would have the adverse effect of perpetuating the knowledge gap about the health condition. Conducting research in emergency situations is an important component of our ability to safely and effectively address the health needs of current and future victims of the emergency situation.

BOX 2-4

Inclusion of Pregnant Women and Children in Clinical Trials.

The Effect of Mortality Rate on Equipoise

During the Ebola epidemic the mortality rate was frequently discussed in deliberations about selecting appropriate trial design. Some stakeholders argued that it would be unethical to randomize patients to a standard-of-care arm, when the current standard of care “does not much affect clinical outcomes and the mortality is as high as 70 percent” (Adebamowo et al., 2014, p. 1423). They argued that in such an environment, “it is problematic to insist on randomizing patients when the intervention arm holds out at least the possibility of benefit,” and they maintained that “ethical arguments are not the same for all levels of risk” (Adebamowo et al., 2014, p. 1423). In contrast, proponents of RCTs countered that there were no data to support the assumption that patients with a life-threatening disease would always choose to use a first-in-human experimental product of unknown safety and efficacy (Nelson et al., 2015).

A basic tenet of clinical research is the concept of equipoise, defined as “genuine uncertainty about whether an untested treatment has benefits or risks that exceed those of conventional care” (Adebamowo et al., 2014). The underlying rationale behind this approach is that because there is no agreement that one of the interventions is superior to the others, it is permissible to allow participants to be allocated at random to receive one or more of these interventions and then to observe, measure, and document the outcome. For some, equipoise breaks down or is not applicable in contexts of extremely high mortality and where available options for care offer little benefit (Adebamowo et al., 2014). Because at the time trial designs were being considered it was estimated that Ebola had a mortality rate of 70 percent or more and it was thought that supportive care offered little benefit, the conclusion was reached by some that it was unethical to randomize participants to an investigational agent or to an arm that provided only standard-of-care treatment measures (Caplan et al., 2015; WHO Ebola Response Team, 2014).

There are several problems with this argument. First, it rests on an assumption about the benefits and risks of investigational agents and the consequences of receiving an appropriate standard of care alone that was not supported by sufficient evidence. Notably, for the investigational therapeutic interventions tested during the Ebola epidemic, the available preclinical data were insufficient to determine that a product was more likely to provide a therapeutic advantage to recipients than to worsen their already fragile condition. Given that most of these products were novel and that failure rates for novel interventions in general are in the range of 90 percent, it seems unreasonable to expect that interventions in the early stages of development would have an appreciable therapeutic advantage, let alone have sufficient efficacy to constitute the desired magic bullet (Dawson, 2015; Hay et al., 2014; Thomas et al., 2015). Additionally, in the early stages of the epidemic, some ETUs were unable to provide patients with basic support, including intravenous (IV) fluids and electrolyte management, and offered only oral fluids (MSF, 2016). It was reasonable to expect that mortality rates in patients who did not receive such supportive care would be higher than in patients who did receive necessary physiological support. This calls into question the stark perception that existed, that Ebola had a uniformly high mortality rate and, therefore, that it was futile to try to improve outcomes with the use of standard supportive treatment.

Second, this view seems to presuppose that desperately ill patients cannot be made clinically worse by the adverse effects of potent therapeutic agents or other modalities. However, this is a morally suspect assumption. Even if we assume that Ebola has a 70 percent mortality rate, being given an investigational agent in the early stages of development might lower this risk, but perhaps as likely it might increase it, thereby reducing the survival rate below 30 percent. Third, this position assumes a greater degree of certainty about relevant factors than is warranted. Overall estimates of mortality from emerging infectious diseases are often uncertain and influenced by many factors. Since subclinical cases are often missed or confused for other conditions, mortality estimates can be biased by the fact that only the sickest patients are properly diagnosed (Lipsitch et al., 2015). In retrospect it is clear that initial assumptions about mortality rates and the shape of the epidemic were incorrect. As the response to Ebola improved, the overall mortality rate in the three high-impact countries progressively dropped over the course of the epidemic, from 61.5 percent in July 2014 to 40.7 percent in July 2015; the mortality rate also differed among the three countries, from a high of 66.6 percent in Guinea to 45.1 percent in Liberia and 30.0 percent in Sierra Leone (Johnston, 2015).

Fourth, the position articulated above treats Ebola as an exceptional case. Sound and socially valuable research often takes place among gravely ill participants involving study designs in which novel agents are compared against standard therapies. Preventing such studies on the grounds that they deny sick patients the chance of receiving a potentially beneficial intervention would create or exacerbate gaps in our knowledge about how best to treat the patients with such conditions. Preventing these studies would reduce our ability to efficiently form an accurate picture of the relative merits and hazards of novel interventions. At the height of the AIDS crisis, before there were any proven treatments, the mortality rate of untreated AIDS was essentially 100 percent. While AIDS differs from Ebola in many ways (e.g., incubation period), similar arguments were made against placebo-controlled randomized trials from AIDS activists such as AIDS Coalition to Unleash Power (ACT UP) (Crimp, 2011; Dawson, 2015; KFF, 2014). In an effort to address their concerns, a group of statisticians (Byar et al., 1990) advocated for randomization in AIDS trials, but also clearly articulated a limited set of conditions for which randomization may not be appropriate. While a universally poor prognosis was one condition, it could not be the only one; rather, Byar et al. argued, all five must apply for uncontrolled trials to be warranted (see Box 2-5).

BOX 2-5

Special Situations in Which Uncontrolled Phase 3 Trials May Be Warranted.

These considerations support the view that equipoise is applicable to emergency contexts and that it is therefore ethically acceptable to offer participants the chance to participate in a trial that begins in and is designed to disturb a state of equipoise (Nelson et al., 2015).

Determining an Ethical Comparator

Debate about clinical trial designs during the Ebola epidemic also focused on what might constitute an ethical comparator for the evaluation of novel interventions. The choice of comparator is important because it provides the benchmark against which novel interventions are assessed. If a sick person receives a novel intervention and his or her condition improves or worsens, the benefit or harm cannot be attributed to the intervention unless we know what would have happened to that person without the intervention. Because we cannot know the answer to this counterfactual directly, we compare the effect of giving a novel intervention to some patients against a comparator group.

Substantial confusion can arise in the discussion of study comparators because some commonly used terms are themselves either misleading or are frequently used in ways that can be confusing. For example, the claim that a novel intervention will be compared against a placebo control is sometimes taken to be synonymous with the claim that it will be compared against “no treatment.” In this instance potential trial participants may believe that those in the comparator arm will not receive any medical care or treatment of any kind. However, this is rarely if ever the case, as both arms will typically receive standard supportive care. Similarly, the statement that a novel intervention is being compared against a placebo is often used to describe two very different situations. The first is what might be called a “placebo-only” comparison, in which members of the investigational arm receive no therapeutic interventions other than the novel intervention being tested and members of the comparator arm receive no therapeutic intervention other than an inert substance that is delivered in the same manner and looks like the novel intervention. In a “placebo-add-on” design, each person in the investigational and comparator arms of the study receives a standardized treatment package as part of the baseline of his or her care and treatment (see Box 2-6 for further discussion on standard care) (Gupta and Verma, 2013). Members of the investigational arm then receive that investigational agent “on top of” this baseline of care, and members of the comparator arm receive a placebo on top of this same baseline package of care.

BOX 2-6

The Standard of Care Owed to Research Participants.

During the Ebola epidemic, some stakeholders objected to a placebo controlled design, unequivocally stating that trials “should not include a placebo: exposed and vulnerable people in Ebola-affected and low-resource settings shouldn't be led to think they are either being treated or protected when they're not” (MSF, 2014b). Objectors to the use of a placebo also argued that it could be unethical or logistically implausible to administer a placebo treatment to such sick patients, for example, “giving 12 to 24 placebo tablets to a vomiting Ebola patient or a 6-hours-lasting placebo infusion to a patient with coagulopathy.”⁶ While RCTs do not require the use of a placebo, placebos are used to facilitate blinding and thereby control for other factors that may influence patient outcomes, such as the conduct of caregivers⁷ (Vickers and de Craen, 2000). The committee determined that such considerations do not warrant the a priori rejection of the use of a placebo but rather should be taken into consideration within the specific context of a trial.

Resistance to studies that would compare a new intervention to a placebo control may have stemmed partly from a misperception that participants in such studies would be denied all care, including such supportive interventions as aggressive rehydration and management of electrolyte abnormalities. Such a “placebo-only” design would have been unethical, but a placebo-add-on design would have been both ethically permissible and scientifically desirable.

No therapeutics trial that was conducted in the three countries in 2014–2015 used a placebo in the standard-of-care study arm (a placebo add-on); rather when a concurrent control was used, for example in the ZMapp trial, it was compared to a standard-of-care arm alone. However, it is possible that the initial discussion of including a placebo polarized the debate about randomized trials. Those who argued against RCTs may have mistakenly believed that patients randomized to the control group would receive no care, rather than the standard of care. In reality, at the beginning of the epidemic, given the capacity of the health care system and infrastructure and the conditions of working in containment in a hastily constructed facility, the standard of care that was generally available was quite limited. However, the clinical trials that used a standard-of-care arm made a concerted effort to provide the best possible supportive care available. For example, the PREVAIL II trial of ZMapp (which had the advantage of starting later when better supportive care was available) had an “optimized-standard-of-care” control arm in which patients received IV fluids, monitoring of electrolytes and key biochemical parameters, and maintenance of oxygenation and blood pressure support and treatment for other infections when they were identified (Davey, 2016). The PREVAIL II trial also demonstrated that individuals would consent to participation in an RCT when it was clearly explained to them. Nelson et al. wrote that “to build trust, all efforts should be made to improve the local standard of care to include early rapid diagnostic testing, the provision of intravenous fluids, and electrolyte management—all of which are known to be effective in reducing the mortality of Ebola. However, the provision of such resources reinforces the need for a concurrent control group, as such interventions are likely to affect mortality” (Nelson et al., 2015) (see Chapter 6 for further discussion on community engagement). While the availability of treatments remained somewhat variable based on trial site, the patients in the standard-of-care arm and the active arm received the same supportive care, with the only difference between the groups being the provision of the investigational medicinal product. Most of the Ebola trials that were conducted provided all trial participants with supportive care, including IV fluids, hemodynamic or electrolyte monitoring or both, and adjunctive medications (e.g., antimalarials, antibiotics) (Dunning et al., 2016a,c; MSF, 2015; PREVAIL II Writing Group, 2016; Sissoko et al., 2016; van Griensven et al., 2016).

Authoritative research ethics guidance documents hold that, as a general rule, research subjects in the control group of a trial of a diagnostic, therapeutic, or preventive intervention should receive an established effective intervention. However, there are some circumstances under which it may be ethically acceptable to compare a novel intervention against a placebo-only comparator. As stated in CIOMS guidelines, such a placebo may be used when

• there is no established effective intervention;
• withholding an established effective intervention would expose subjects to, at most, temporary discomfort or delay in relief of symptoms; and
• use of an established effective intervention as comparator would not yield scientifically reliable results and use of placebo would not add any risk of serious or irreversible harm to the subjects (CIOMS, 2016).

Guideline 33 of the Declaration of Helsinki (2013) also states that the benefits, risks, burdens, and effectiveness of a new intervention must be tested against those of the best proven intervention(s), except in the following circumstances:

• Where no proven intervention exists, the use of placebo, or no intervention, is acceptable; or
• Where for compelling and scientifically sound methodological reasons the use of any intervention less effective than the best proven one, the use of placebo, or no intervention is necessary to determine the efficacy or safety of an intervention and the patients who receive any intervention less effective than the best proven one, placebo, or no intervention will not be subject to additional risks of serious or irreversible harm as a result of not receiving the best proven intervention. Extreme care must be taken to avoid abuse of this option (WMA, 2013).

Given the severity of Ebola and the likely effects of leaving severe dehydration untreated, a placebo-only trial would not meet the above exceptions and, therefore, would not have been ethically acceptable. Because a placebo-add-on design would not deny any study participant access to currently accepted treatment for Ebola, such comparator arms would be ethically permissible so long as the provision of the placebo add-on would be feasible and could be performed safely. The committee determined that testing a novel intervention against a standard-of-care comparator without a placebo add-on is less desirable from a methodological point of view, but is also ethically acceptable, particularly where administering a placebo involves risk to providers or patients (e.g., because of the difficulty of injecting highly infectious patients). Whether IV fluids are actually provided is another issue. As Lamontagne et al. observed, “A common assumption is that a lack of material resources constitutes the dominant barrier to clinical care. That is not the case. Intravenous catheters, fluids, and electrolyte replacement are readily available but thus far are being used much too sparingly. . . . There is a historical bias against aggressive interventions, including intravenous cannulation, for many transmissible illnesses. Percutaneous injury to health care workers does carry substantial risk, but such risks are not specific to Ebola” (Lamontagne et al., 2014, p. 1566).

On this point, it is worth considering the implications of a standard of care or placebo control group with some level of the risk of serious harm to subjects, researchers, or caregivers. For some therapies, for example, using IVs for rehydration or administering a placebo in infected people presented risks to caregivers, as administering an IV involves potential risk from an accidental needle stick to the person doing the infusion. However, as it became common to provide IV fluids as part of standard supportive care the evidence suggests the marginal increase in risk was modest. In March 2015 Partners In Health stated that “responders started putting IVs in children more regularly to resuscitate and rehydrate them. In some cases, they used intraosseous lines (inserted into the bone) if insertion into a vein wasn't possible. . . . The more aggressive use of hydration has certainly dropped mortality rates” (Partners In Health, 2015). The result was that IV rehydration became standard care for all patients at the Maforki Ebola Treatment Unit in Port Loko, Sierra Leone. In areas where IV rehydration was already the standard of care and the health care team had experience, the additional risk to researchers and caregivers to administer an IV placebo, if deemed necessary, would be limited, although needle-stick injury is a well-described hazard for transmission of Ebola (Guardian, 2015). The greater exposure risk may actually result from the more extensive bedside monitoring and direct contact with patients or with the equipment used for supportive care, such as respirators. During the Ebola-TKM trial, monitoring occurred over the 2-hour infusion period, with additional assessment of vital signs before, during, and at 1, 2, 4, and 8 hours after the end of infusion (Dunning et al., 2016c). For these reasons the committee determined that the additional risk involved with administering an IV placebo is minimal because IV fluid replacement is considered standard of care for Ebola and all patients with dehydration should be receiving fluids, whether they are in the treatment or the control arm of a study.

It might, however, be challenging to allocate the manpower required to monitor a patient receiving a placebo or IV for an extended period of time. It can be argued that unless researchers are able to provide all of the necessary resources to conduct a trial properly and not interrupt clinical care routines, they ought not to proceed. A call for more volunteers to carry out protocol requirements would be relevant, but it may also be insensitive or naïve to simply call for more manpower. The priority is to set up enough facilities and staff to deliver a site-specific optimized standard of care. With that in place, RCTs are more palatable and readily implemented.

In principle, there is the danger that providing an enhanced standard of care even in the control arm might push people to enroll in trials (McMillan and Conlon, 2004); however, it is unclear whether this was a factor in community members' decisions to enroll in Ebola trials.

Limited Product Availability

Supplies of some of the experimental therapies being considered for Ebola were limited, and this influenced perspectives on the appropriateness of RCTs under these circumstances. Objectors to RCTs believed that RCTs would deprive some patients of access to treatment. Proponents of RCTs countered that “given the scarcity of the drug, a finite number of patients will receive access regardless of what study design is used” (Joffe, 2014, p. 1300), and they maintained that RCTs would actually be a fair and ethical way to allocate resources while gathering data that could help future Ebola patients (Cox et al., 2014). Further, supporters of randomization noted that “alternative means for prioritizing access, such as first-come first-served and sickest first, are themselves ethically unsatisfactory” (Joffe, 2014, p. 1300) and would fail to generate interpretable evidence.

The argument for randomization seems especially strong in the case of a limited supply of product. If there were, for example, just 10 doses of an experimental therapy available, would it be rational to give it to the first 10 people who showed up and agreed to its administration when this may sacrifice the ability to learn something about its efficacy and safety? Would the decision be different if there were 100 doses? In contrast, if an intervention is in surplus and randomization is applied, it might appear to some observers that the researchers are “withholding” the intervention from some patients. All trial designs need to enroll enough subjects to reach interpretable endpoints and there needs to be enough available product for the patients randomized to the experimental treatment. However, even small randomized studies can provide a provisional assessment of efficacy that a first-come/first-served approach cannot, unless the effect is very dramatic. If not enough doses to conduct a suitable trial are available, the use of a randomized lottery system to distribute the available doses would still be preferential because it would be the fairest possible way to distribute scarce resources and retain the potential to generate useful information. The Ebola report of the Presidential Commission for the Study of Bioethical Issues presented two competing perspectives on obligations to Africans confronting Ebola (Presidential Commission for the Study of Bioethical Issues, 2015):

1.

Identify safe, effective interventions as efficiently and reliably as possible.

2.

Provide access to the potential benefits of experimental interventions to as many people as possible using scientifically valid research designs.

For scarce interventions during the epidemic, like ZMapp, there may be no conflict between these two obligations. But where the intervention may be given to a large number of people (e.g., vaccines), there may be controversy over a randomized trial because there must be a risk of infection distributed across the population to be able to assess efficacy, and some subjects will be denied the potential benefit of the product. On the other hand, they are spared the potential adverse effects of the vaccine. Many still recall the widespread administration of an experimental vaccine in the United States to protect against the emerging influenza A H1N1 New Jersey 1976 strain that provided no protection because the anticipated epidemic did not happen; however, there was a sharp increase in the number of individuals with Guillain-Barre syndrome in the weeks following immunization, and 25 deaths were attributed to the vaccine (Langmuir, 1979).

Community Trust

Some argued that mistrust of health care workers by the community would deter trial enrollment and the generation of meaningful data (CIOMS Guideline 1) (Caplan et al., 2015). They also argued that the sharing of drugs (to ensure that more people had access to the active agent, as seen in early HIV trials with AZT [Farber, 2015]) might compromise trial results. However, previous experiences in confronting the HIV epidemic revealed that controlled conditions of a randomized trial are possible even in developing country settings, and even where there is a sense of urgency about addressing an emerging epidemic (Lane et al., 2016). These concerns, in combination with the violence and mistrust the affected population had shown toward both ETUs and health care workers, led some to assume that randomization would be locally unacceptable (Adebamowo et al., 2014; McCoy, 2014). Waldman and Neiburg expanded on this, emphasizing that the confusion and volatility of the situation would make it challenging for individuals to understand what the standard of care was (Waldman and Nieburg, 2015). They worried that the foreign care providers would be perceived as providing potentially lifesaving treatments, thus enhancing therapeutic misperceptions. At the very least, this situation would make fully informed consent challenging. “RCTs will not work without community trust,” Caplan et al. wrote, “yet implementing them risks eroding that trust” (Caplan et al., 2015, p. 7).

It is critically important, however, that researchers not make assumptions about how communities perceive prospective research activities. In order to lay the groundwork for ethical research, there has to be truthful engagement with the community about the tradeoffs inherent to clinical trials so they can make an informed decision about the trial designs that can be implemented. If trial teams rush to enter a community in order to rapidly implement a trial without having a proper engagement strategy, it can backfire, as observed during the latter months of 2014. When the teams can engage the community and provide clearly articulated information about the various aspects of trial designs, community buy-in for research is possible, as documented by the experience in West Africa. For example, the PREVAIL team demonstrated their ability to propose and implement an RCT in Liberia in early 2015 (Wilson et al., 2016). Respect for communities requires that their members are engaged in a process of dialogue and exchange with the investigators about the need for research, the nature of the uncertainty to be addressed, what is known about the status of the interventions to be used, and the merits of possible trial designs. As the committee heard through testimony in their meeting in Liberia, communities, even those that were previously unexposed to clinical trials, are capable of understanding components of research when it is explained. In a context of scarcity, need, and heightened mistrust, such conversations can be challenging. But they are an indispensable component of ethically sound research and are critical to treating study communities as full partners in the effort to find the means to advance their health needs.

However, if after substantial and genuine community discussion and engagement there is still extensive opposition to trials that involve randomization, it becomes reasonable to consider alternative trial approaches. When this course is taken there must be a commitment to avoid supporting any design that is unlikely to produce sufficiently reliable evidence in order to offset the many risks, costs, and burdens associated with research. The study design proposed by Cooper et al., using a single-arm study at the outset and then moving to a randomized trial if the results are promising but neither exceptional nor limited, is an effort to address the concerns of such communities while recognizing that views about the acceptability of randomization may evolve as evidence accumulates and it is clear that an intervention is or is not highly efficacious (Cooper et al., 2015). Such multistage study designs may represent prudent options in such circumstances, when despite engagement and information exchanges, communities or the health authorities in a country will not accept the inclusion of an add-on control group in a clinical trial. Research must be conducted in a responsible and locally acceptable fashion, with attention paid to the local communities' values, beliefs, and priorities. Failing to conduct research in this manner risks more than the success of the research project; it can also jeopardize the trust and relationships that allow clinical care to be delivered to the community. (See Chapter 6 for further discussion of community engagement.)