The goal of a typical phase I trial is to document the safety of a new intervention, most commonly by determining the dose-limiting toxicity (DLT) and maximum tolerated dose (MTD) of a novel drug. Critics contend that a phase I trial’s objective is to learn about safety, without therapeutic intent, raising the important question – can phase I trials offer the prospect of benefit to a child with cancer? (Miller and Joffe 2008; Agrawal and Emanuel 2003). By virtue of the animal data required to justify a phase I trial, the drug must show promise for patient subjects. And indeed, some studies suggest approximately 10 % of subjects have a partial or complete response (Lee et al. 2005). Moreover, many investigators insist that participants receive psychological benefits by contributing to future knowledge and by maintaining hope (Kodish et al. 1992). Once we accept that a trial offers the prospect of direct benefit, it must satisfy other regulatory criteria: (1) the risks of exposing a child to a novel intervention is justified by the anticipated benefit, and (2) the relation of anticipated benefit to risk is at least as favorable as that available with alternative approaches (Office for Protection from Research Risks 1983).

The final ethical concern we will address is that of informed consent. Critics contend that parents’ desperation for a cure makes them vulnerable. Available data demonstrates that participants enroll in phase I trials largely for personal benefit (Daugherty et al. 2000), that they overestimate the likelihood of benefit (Miller 2000), and that they have a poor understanding of the primary dose-finding objective of these studies (Cousino et al. 2012). While there is a dearth of quality data to attribute this to inadequate disclosure or instead to optimism, it is clear that a requisite component of any phase I trial is a transparent and forthright consent process.

Randomized controlled trials (RCTs) are the gold standard of biomedical research, allowing investigators to directly compare two interventions while controlling for factors that confound data generated by other study designs. In a typical oncology RCT, each subject is assigned, by chance, to one of two study groups. Subjects in the control group receive standard therapy, while those in the intervention group receive the treatment of interest. The classic ethical justification for such studies revolves around the concept of equipoise. Equipoise refers to “a state of professional uncertainty about [the] relative therapeutic merits” of the two interventions being studied (Miller and Joffe 2011). Put simply, if one of the two options is known to be better, we cannot justify randomly condemning half of the study population to receive the lesser therapy. Moreover, there is no need to conduct the study if we are certain which treatment is better. Conversely, most investigational therapies do not make it to a RCT unless there is evidence to suggest efficacy, so for a disease with poor therapeutic options, the presence of promising preliminary data might move expert opinion in favor of the novel drug, despite the lack of rigorous supporting data.

There are a host of criticisms of the concept of equipoise, and its use as the rationale for the appropriateness of RCTs (Miller and Brody 2003). Firstly, most patients expect their physician to weigh evidence, patient characteristics, and physician experience to determine the best therapeutic option. Equipoise assumes that patients will accept having that decision randomly assigned instead. Additionally, patients themselves may have a preference between treatments, even if the medical community does not. Thirdly, once a study begins, there may reach a point at which preliminary data favors one treatment over another. At this point equipoise, by definition, would be disturbed, and continuing the trial would be considered unethical. Data-monitoring committees are a part of most RCTs, and based on interim, data must either allow a study to continue or halt the study based on predefined stopping rules.

The National Bioethics Advisory Commission defines the therapeutic misconception as “the belief that the purpose of a clinical trial is to benefit the individual patient rather than to gather data for the purpose of contributing to scientific knowledge” (National Bioethics Advisory Commission 2001). This concept is similar to the discussion of informed consent in phase I trials; however, it is also pertinent to RCTs such as the one described above. Despite clinicians’ best intentions, a significant percent of parents choosing the RCT will mistakenly believe that the clinician will select the treatment based on her judgment of which option is the best fit for the child (Kodish et al. 2004). There is increasing evidence that this type of misunderstanding is widespread, leading to calls for new approaches to trial enrollment, such as preventing clinicians from being the ones to present study details and offer enrollment (Eder et al. 2007; Appelbaum et al. 2012; Flory and Emanuel 2004).

Health-care providers have an undeniable moral duty to treat the pain and suffering of terminally ill patients. Yet this imperative may also raise ethical questions if treating the pain and suffering could possibly hasten a patient’s death. Although empirical evidence suggests that the use of opioids and sedatives to relieve symptoms at the end of life in no way shortens life, theoretical concern remains (Macauley 2012; López-Saca et al. 2013). The ethical principle relevant to this question is the Doctrine of Double Effect (May 1978). The Doctrine states that when an action has two effects, one of which is inherently good and the other inherently bad, it can be justified if certain conditions are met:

The moral reasoning outlined above remains controversial for a few reasons. First, according to the Doctrine, the moral justification for a clinician’s practice hinges on his intention, something that is impossible to know with certainty. In addition, critics believe that the only morally relevant consideration is the informed consent of the patient (or appropriate proxy), which is not explicitly considered in the Doctrine (Quill et al. 1997; Truog et al. 2012).

A rarely used approach to sedation and analgesia at the end of life is palliative sedation, wherein a child must be sedated to the point of unconsciousness in order to be comfortable (Truog et al. 1992; Cowan and Walsh 2001; Postovsky et al. 2007). This practice should be considered, with consent, when all other methods of controlling the patient’s suffering have failed (National Ethics Committee 2006). After the patient loses consciousness, hydration and nutrition stop and no attempt is typically made to restore the patient to consciousness. Patients typically die after several days from dehydration or a complication of the treatment.

The US Supreme Court implicitly endorsed palliative sedation, citing the technique as an alternative to physician-assisted suicide that could, in theory, assure that no patient should die with “untreatable” pain (McStay 2003). At least in part because of this legal endorsement, palliative sedation has become more widely practiced, although its ethical justification remains controversial (Schüklenk et al. 2011; Kiman et al. 2011; Goldstein et al. 2012).

The decision to withhold or withdraw medical nutrition and hydration should be made after weighing the burdens and benefits of therapy for that patient. Such decisions are morally equivalent to those pertaining to mechanical ventilation and other life-sustaining interventions (Levi 2003; Diekema and Botkin 2009; Nelson 1987); however, they can be controversial since many consider nutrition and hydration to be basic supports for ongoing life, without which human dignity is compromised. Some argue that decisions to forego medical nutrition and hydration can constitute patient abandonment. Despite these concerns, courts have ruled that they are medical interventions and may be discontinued on the same grounds as any other medical treatment (Table 17.3).